bump to 0.6.2

add new boards
add abort command
2024-11-01 15:19:37 +08:00 · 2024-11-01 14:26:02 +08:00 · 2024-10-31 05:57:13 +08:00 · 2024-10-30 06:58:29 +08:00 · 2024-10-29 00:26:05 +08:00 · 2024-10-29 00:22:29 +08:00
56 changed files with 2328 additions and 1007 deletions
--- a/.gitignore
+++ b/.gitignore
@@ -8,3 +8,4 @@ sdkconfig.old
 sdkconfig
 dependencies.lock
 .env
+releases/
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -4,7 +4,7 @@
 # CMakeLists in this exact order for cmake to work correctly
 cmake_minimum_required(VERSION 3.16)

-set(PROJECT_VER "0.3.3")
+set(PROJECT_VER "0.6.2")

 include($ENV{IDF_PATH}/tools/cmake/project.cmake)
 project(xiaozhi)
--- a/README.md
+++ b/README.md
@@ -1,9 +1,11 @@
 # 小智 AI 聊天机器人

-BiliBili 视频介绍 [【ESP32+SenseVoice+Qwen72B打造你的AI聊天伴侣！】](https://www.bilibili.com/video/BV11msTenEH3/?share_source=copy_web&vd_source=ee1aafe19d6e60cf22e60a93881faeba)
-
 这是虾哥的第一个硬件作品。

+[ESP32+SenseVoice+Qwen72B打造你的AI聊天伴侣！【bilibili】](https://www.bilibili.com/video/BV11msTenEH3/?share_source=copy_web&vd_source=ee1aafe19d6e60cf22e60a93881faeba)
+
+[手工打造你的 AI 女友，新手入门教程【bilibili】](https://www.bilibili.com/video/BV1XnmFYLEJN/)
+
 ## 项目目的

 本项目基于乐鑫的 ESP-IDF 进行开发。
@@ -18,16 +20,16 @@ BiliBili 视频介绍 [【ESP32+SenseVoice+Qwen72B打造你的AI聊天伴侣！

 - Wi-Fi 配网
 - 支持 BOOT 键唤醒和打断
- 离线语音唤醒（使用乐鑫方案）
+- 离线语音唤醒（乐鑫方案）
 - 流式语音对话（WebSocket 协议）
- 支持国语、粤语、英语、日语、韩语 5 种语言识别（使用 SenseVoice 方案）
+- 支持国语、粤语、英语、日语、韩语 5 种语言识别（SenseVoice 方案）
 - 声纹识别（识别是谁在喊 AI 的名字，[3D Speaker 项目](https://github.com/modelscope/3D-Speaker)）
- 使用大模型 TTS（火山引擎方案，阿里云接入中）
+- 使用大模型 TTS（火山引擎与 CosyVoice 方案）
 - 支持可配置的提示词和音色（自定义角色）
- 免费提供 Qwen2.5 72B 和 豆包模型（受限于性能和额度，人多后可能会限额）
+- Qwen2.5 72B 或 豆包 API
 - 支持每轮对话后自我总结，生成记忆体
- 扩展液晶显示屏，显示信号强弱（后面可以显示中文字幕）
- 支持 ML307 Cat.1 4G 模块（可选）
+- 扩展液晶显示屏，显示信号强弱
+- 支持 ML307 Cat.1 4G 模块

 ## 硬件部分

@@ -35,60 +37,29 @@ BiliBili 视频介绍 [【ESP32+SenseVoice+Qwen72B打造你的AI聊天伴侣！

 [《小智 AI 聊天机器人百科全书》](https://ccnphfhqs21z.feishu.cn/wiki/F5krwD16viZoF0kKkvDcrZNYnhb?from=from_copylink)

-第二版接线图如下：
+面包板接线图如下：

-![第二版接线图](docs/wiring2.jpg)
+![面包板接线图](docs/wiring2.jpg)

 ## 固件部分

 ### 免开发环境烧录

-新手第一次操作建议先不要搭建开发环境，直接使用免开发环境烧录的固件。
+新手第一次操作建议先不要搭建开发环境，直接使用免开发环境烧录的固件。固件使用的是作者友情提供的测试服，目前开放免费使用，请勿用于商业用途。

-点击 [这里](https://github.com/78/xiaozhi-esp32/releases) 下载最新版固件。
+[Flash烧录固件（无IDF开发环境）](https://ccnphfhqs21z.feishu.cn/wiki/Zpz4wXBtdimBrLk25WdcXzxcnNS) 

-固件使用的是作者友情提供的测试服，目前开放免费使用，请勿用于商业用途。

-### 搭建开发环境
+### 开发环境

 - Cursor 或 VSCode
 - 安装 ESP-IDF 插件，选择 SDK 版本 5.3 或以上
 - Ubuntu 比 Windows 更好，编译速度快，也免去驱动问题的困扰

-### 配置项目与编译固件

- 目前只支持 ESP32 S3，Flash 至少 8MB, PSRAM 至少 2MB（注意：默认配置只兼容 8MB PSRAM，如果你使用 2MB PSRAM，需要修改配置，否则无法识别）
- 配置 OTA Version URL 为 `https://api.tenclass.net/xiaozhi/ota/`
- 配置 WebSocket URL 为 `wss://api.tenclass.net/xiaozhi/v1/`
- 配置 WebSocket Access Token 为 `test-token`
- 如果 INMP441 和 MAX98357 接线跟默认配置不一样，需要修改 GPIO 配置
- 配置完成后，编译固件
+## AI 角色配置

+如果你已经拥有一个小智 AI 聊天机器人，可以参考 [后台操作视频教程](https://www.bilibili.com/video/BV1jUCUY2EKM/)

-## 配置 Wi-Fi （4G 版本跳过）
-
-按照上述接线，烧录固件，设备上电后，开发板上的 RGB 会闪烁蓝灯（部分开发板需要焊接 RGB 灯的开关才会亮），进入配网状态。
-
-打开手机 Wi-Fi，连接上设备热点 `Xiaozhi-xxxx` 后，使用浏览器访问 `http://192.168.4.1`，进入配网页面。
-
-选择你的路由器 WiFi，输入密码，点击连接，设备会在 3 秒后自动重启，之后设备会自动连接到路由器。
-
-## 测试设备是否连接成功
-
-设备连接上路由器后，闪烁一下绿灯。此时，喊一声“你好，小智”，设备会先亮蓝灯（表示连接服务器），然后再亮绿灯，播放语音。
-
-如果没有亮蓝灯，说明麦克风有问题，请检查接线是否正确。
-
-如果没有亮绿灯，或者蓝灯常亮，说明设备没有连接到服务器，请检查 WiFi 连接是否正常。
-
-如果设备已经连接 Wi-Fi，但是没有声音，请检查是否接线正确。
-
-在 v0.2.1 版本之后的固件，也可以按下连接 GPIO 1 的按钮（低电平有效），进行录音测试。
-
-## 配置设备
-
-如果上述步骤测试成功，设备会播报你的设备 ID，你需要到 [小智测试服的控制面板](https://xiaozhi.tenclass.net/) 页面，添加设备。
-
-详细的使用说明以及测试服的注意事项，请参考 [小智测试服的帮助说明](https://xiaozhi.tenclass.net/help)。
-
+详细的使用说明以及测试服的注意事项，请参考 [小智测试服的帮助说明](https://xiaozhi.me/help)。

--- a/convert_audio_to_p3.py
+++ b/convert_audio_to_p3.py
@@ -0,0 +1,40 @@
+# convert audio files to protocol v3 stream
+import librosa
+import opuslib
+import struct
+import sys
+import tqdm
+import numpy as np
+
+def encode_audio_to_opus(input_file, output_file):
+    # Load audio file using librosa
+    audio, sample_rate = librosa.load(input_file, sr=None, mono=False, dtype=np.int16)
+    
+    # Get left channel if stereo
+    if audio.ndim == 2:
+        audio = audio[0]
+    
+    # Initialize Opus encoder
+    encoder = opuslib.Encoder(sample_rate, 1, opuslib.APPLICATION_VOIP)
+
+    # Encode audio data to Opus packets
+    # Save encoded data to file
+    with open(output_file, 'wb') as f:
+        sample_rate = 16000 # 16000Hz
+        duration = 60 # 60ms every frame
+        frame_size = int(sample_rate * duration / 1000)
+        for i in tqdm.tqdm(range(0, len(audio) - frame_size, frame_size)):
+            frame = audio[i:i + frame_size]
+            opus_data = encoder.encode(frame.tobytes(), frame_size=frame_size)
+            # protocol format, [1u type, 1u reserved, 2u len, data]
+            packet = struct.pack('>BBH', 0, 0, len(opus_data)) + opus_data
+            f.write(packet)
+
+# Example usage
+if len(sys.argv) != 3:
+    print('Usage: python convert.py <input_file> <output_file>')
+    sys.exit(1)
+
+input_file = sys.argv[1]
+output_file = sys.argv[2]
+encode_audio_to_opus(input_file, output_file)
--- a/docs/wiring2.jpg
+++ b/docs/wiring2.jpg
--- a/main/Application.cc
+++ b/main/Application.cc
@@ -1,6 +1,4 @@
 #include <BuiltinLed.h>
-#include <TcpTransport.h>
-#include <TlsTransport.h>
 #include <Ml307SslTransport.h>
 #include <WifiConfigurationAp.h>
 #include <WifiStation.h>
@@ -10,46 +8,60 @@
 #include <esp_log.h>
 #include <cJSON.h>
 #include <driver/gpio.h>
+#include <arpa/inet.h>

 #include "Application.h"

 #define TAG "Application"

+extern const char p3_err_reg_start[] asm("_binary_err_reg_p3_start");
+extern const char p3_err_reg_end[] asm("_binary_err_reg_p3_end");
+extern const char p3_err_pin_start[] asm("_binary_err_pin_p3_start");
+extern const char p3_err_pin_end[] asm("_binary_err_pin_p3_end");
+extern const char p3_err_wificonfig_start[] asm("_binary_err_wificonfig_p3_start");
+extern const char p3_err_wificonfig_end[] asm("_binary_err_wificonfig_p3_end");
+

 Application::Application()
-    : boot_button_((gpio_num_t)CONFIG_BOOT_BUTTON_GPIO),
-      volume_up_button_((gpio_num_t)CONFIG_VOLUME_UP_BUTTON_GPIO)
-#ifdef CONFIG_USE_ML307
-    , ml307_at_modem_(CONFIG_ML307_TX_PIN, CONFIG_ML307_RX_PIN, 4096),
-      http_(ml307_at_modem_),
-      firmware_upgrade_(http_)
-#else
-    , http_(),
-    firmware_upgrade_(http_)
-#endif
-#ifdef CONFIG_USE_DISPLAY
-    , display_(CONFIG_DISPLAY_SDA_PIN, CONFIG_DISPLAY_SCL_PIN)
-#endif
+    : boot_button_((gpio_num_t)BOOT_BUTTON_GPIO),
+      volume_up_button_((gpio_num_t)VOLUME_UP_BUTTON_GPIO),
+      volume_down_button_((gpio_num_t)VOLUME_DOWN_BUTTON_GPIO),
+      display_(DISPLAY_SDA_PIN, DISPLAY_SCL_PIN)
 {
    event_group_ = xEventGroupCreate();
-    
-    opus_encoder_.Configure(CONFIG_AUDIO_INPUT_SAMPLE_RATE, 1);
+
+    opus_encoder_.Configure(16000, 1);
    opus_decoder_ = opus_decoder_create(opus_decode_sample_rate_, 1, NULL);
-    if (opus_decode_sample_rate_ != CONFIG_AUDIO_OUTPUT_SAMPLE_RATE) {
-        opus_resampler_.Configure(opus_decode_sample_rate_, CONFIG_AUDIO_OUTPUT_SAMPLE_RATE);
+    if (opus_decode_sample_rate_ != AUDIO_OUTPUT_SAMPLE_RATE) {
+        output_resampler_.Configure(AUDIO_OUTPUT_SAMPLE_RATE, opus_decode_sample_rate_);
+    }
+    if (16000 != AUDIO_INPUT_SAMPLE_RATE) {
+        input_resampler_.Configure(AUDIO_INPUT_SAMPLE_RATE, 16000);
    }

    firmware_upgrade_.SetCheckVersionUrl(CONFIG_OTA_VERSION_URL);
    firmware_upgrade_.SetHeader("Device-Id", SystemInfo::GetMacAddress().c_str());
-    firmware_upgrade_.SetPostData(SystemInfo::GetJsonString());
 }

 Application::~Application() {
+    if (update_display_timer_ != nullptr) {
+        esp_timer_stop(update_display_timer_);
+        esp_timer_delete(update_display_timer_);
+    }
+    if (ws_client_ != nullptr) {
+        delete ws_client_;
+    }
    if (opus_decoder_ != nullptr) {
        opus_decoder_destroy(opus_decoder_);
    }
    if (audio_encode_task_stack_ != nullptr) {
-        free(audio_encode_task_stack_);
+        heap_caps_free(audio_encode_task_stack_);
+    }
+    if (main_loop_task_stack_ != nullptr) {
+        heap_caps_free(main_loop_task_stack_);
+    }
+    if (audio_device_ != nullptr) {
+        delete audio_device_;
    }

    vEventGroupDelete(event_group_);
@@ -57,6 +69,7 @@ Application::~Application() {

 void Application::CheckNewVersion() {
    // Check if there is a new firmware version available
+    firmware_upgrade_.SetBoardJson(Board::GetInstance().GetJson());
    firmware_upgrade_.CheckVersion();
    if (firmware_upgrade_.HasNewVersion()) {
        // Wait for the chat state to be idle
@@ -65,11 +78,9 @@ void Application::CheckNewVersion() {
        }
        SetChatState(kChatStateUpgrading);
        firmware_upgrade_.StartUpgrade([this](int progress, size_t speed) {
-#ifdef CONFIG_USE_DISPLAY
            char buffer[64];
            snprintf(buffer, sizeof(buffer), "Upgrading...\n %d%% %zuKB/s", progress, speed / 1024);
            display_.SetText(buffer);
-#endif
        });
        // If upgrade success, the device will reboot and never reach here
        ESP_LOGI(TAG, "Firmware upgrade failed...");
@@ -79,138 +90,98 @@ void Application::CheckNewVersion() {
    }
 }

-#ifdef CONFIG_USE_DISPLAY
+void Application::Alert(const std::string&& title, const std::string&& message) {
+    ESP_LOGE(TAG, "Alert: %s, %s", title.c_str(), message.c_str());
+    display_.ShowNotification(std::string(title + "\n" + message));

-#ifdef CONFIG_USE_ML307
-static std::string csq_to_string(int csq) {
-    if (csq == -1) {
-        return "No network";
-    } else if (csq >= 0 && csq <= 9) {
-        return "Very bad";
-    } else if (csq >= 10 && csq <= 14) {
-        return "Bad";
-    } else if (csq >= 15 && csq <= 19) {
-        return "Fair";
-    } else if (csq >= 20 && csq <= 24) {
-        return "Good";
-    } else if (csq >= 25 && csq <= 31) {
-        return "Very good";
-    }
-    return "Invalid";
-}
-#else
-static std::string rssi_to_string(int rssi) {
-    if (rssi >= -55) {
-        return "Very good";
-    } else if (rssi >= -65) {
-        return "Good";
-    } else if (rssi >= -75) {
-        return "Fair";
-    } else if (rssi >= -85) {
-        return "Poor";
-    } else {
-        return "No network";
+    if (message == "PIN is not ready") {
+        PlayLocalFile(p3_err_pin_start, p3_err_pin_end - p3_err_pin_start);
+    } else if (message == "Configuring WiFi") {
+        PlayLocalFile(p3_err_wificonfig_start, p3_err_wificonfig_end - p3_err_wificonfig_start);
+    } else if (message == "Registration denied") {
+        PlayLocalFile(p3_err_reg_start, p3_err_reg_end - p3_err_reg_start);
    }
 }
-#endif

-void Application::UpdateDisplay() {
-    while (true) {
-        if (chat_state_ == kChatStateIdle) {
-#ifdef CONFIG_USE_ML307
-            std::string network_name = ml307_at_modem_.GetCarrierName();
-            int signal_quality = ml307_at_modem_.GetCsq();
-            if (signal_quality == -1) {
-                network_name = "No network";
-            } else {
-                ESP_LOGI(TAG, "%s CSQ: %d", network_name.c_str(), signal_quality);
-                display_.SetText(network_name + "\n" + csq_to_string(signal_quality) + " (" + std::to_string(signal_quality) + ")");
-            }
-#else
-            auto& wifi_station = WifiStation::GetInstance();
-            int8_t rssi = wifi_station.GetRssi();
-            display_.SetText(wifi_station.GetSsid() + "\n" + rssi_to_string(rssi) + " (" + std::to_string(rssi) + ")");
-#endif
-        }
-        vTaskDelay(pdMS_TO_TICKS(10 * 1000));
+void Application::PlayLocalFile(const char* data, size_t size) {
+    ESP_LOGI(TAG, "PlayLocalFile: %zu bytes", size);
+    SetDecodeSampleRate(16000);
+
+    {
+        std::lock_guard<std::mutex> lock(mutex_);
+        auto packet = new AudioPacket();
+        packet->type = kAudioPacketTypeStart;
+        audio_decode_queue_.push_back(packet);
+    }
+
+    ParseBinaryProtocol3(data, size);
+
+    {
+        std::lock_guard<std::mutex> lock(mutex_);
+        auto packet = new AudioPacket();
+        packet->type = kAudioPacketTypeStop;
+        audio_decode_queue_.push_back(packet);
+        cv_.notify_all();
    }
 }
-#endif

 void Application::Start() {
    auto& builtin_led = BuiltinLed::GetInstance();
-#ifdef CONFIG_USE_ML307
    builtin_led.SetBlue();
    builtin_led.StartContinuousBlink(100);
-    ml307_at_modem_.SetDebug(false);
-    ml307_at_modem_.SetBaudRate(921600);
-    // Print the ML307 modem information
-    std::string module_name = ml307_at_modem_.GetModuleName();
-    ESP_LOGI(TAG, "ML307 Module: %s", module_name.c_str());
-#ifdef CONFIG_USE_DISPLAY
-    display_.SetText(std::string("Wait for network\n") + module_name);
-#endif
-    ml307_at_modem_.ResetConnections();
-    ml307_at_modem_.WaitForNetworkReady();

-    ESP_LOGI(TAG, "ML307 IMEI: %s", ml307_at_modem_.GetImei().c_str());
-    ESP_LOGI(TAG, "ML307 ICCID: %s", ml307_at_modem_.GetIccid().c_str());
+    auto& board = Board::GetInstance();
+    board.Initialize();

-    // If low power, the material ready event will be triggered by the modem because of a reset
-    ml307_at_modem_.OnMaterialReady([this]() {
-        ESP_LOGI(TAG, "ML307 material ready");
-        Schedule([this]() {
-            SetChatState(kChatStateIdle);
-        });
-    });
-#else
-    // Try to connect to WiFi, if failed, launch the WiFi configuration AP
-    auto& wifi_station = WifiStation::GetInstance();    
-#ifdef CONFIG_USE_DISPLAY
-    display_.SetText(std::string("Connect to WiFi\n") + wifi_station.GetSsid());
-#endif
-    builtin_led.SetBlue();
-    builtin_led.StartContinuousBlink(100);
-    wifi_station.Start();
-    if (!wifi_station.IsConnected()) {
-        builtin_led.SetBlue();
-        builtin_led.Blink(1000, 500);
-        auto& wifi_ap = WifiConfigurationAp::GetInstance();
-        wifi_ap.SetSsidPrefix("Xiaozhi");
-#ifdef CONFIG_USE_DISPLAY
-        display_.SetText(wifi_ap.GetSsid() + "\n" + wifi_ap.GetWebServerUrl());
-#endif
-        wifi_ap.Start();
-        return;
-    }
-#endif
-
-    audio_device_.OnInputData([this](const int16_t* data, int size) {
+    audio_device_ = board.CreateAudioDevice();
+    audio_device_->Initialize();
+    audio_device_->EnableOutput(true);
+    audio_device_->EnableInput(true);
+    audio_device_->OnInputData([this](std::vector<int16_t>&& data) {
+        if (16000 != AUDIO_INPUT_SAMPLE_RATE) {
+            if (audio_device_->input_channels() == 2) {
+                auto left_channel = std::vector<int16_t>(data.size() / 2);
+                auto right_channel = std::vector<int16_t>(data.size() / 2);
+                for (size_t i = 0, j = 0; i < left_channel.size(); ++i, j += 2) {
+                    left_channel[i] = data[j];
+                    right_channel[i] = data[j + 1];
+                }
+                auto resampled_left = std::vector<int16_t>(input_resampler_.GetOutputSamples(left_channel.size()));
+                auto resampled_right = std::vector<int16_t>(input_resampler_.GetOutputSamples(right_channel.size()));
+                input_resampler_.Process(left_channel.data(), left_channel.size(), resampled_left.data());
+                input_resampler_.Process(right_channel.data(), right_channel.size(), resampled_right.data());
+                data.resize(resampled_left.size() + resampled_right.size());
+                for (size_t i = 0, j = 0; i < resampled_left.size(); ++i, j += 2) {
+                    data[j] = resampled_left[i];
+                    data[j + 1] = resampled_right[i];
+                }
+            } else {
+                auto resampled = std::vector<int16_t>(input_resampler_.GetOutputSamples(data.size()));
+                input_resampler_.Process(data.data(), data.size(), resampled.data());
+                data = std::move(resampled);
+            }
+        }
 #ifdef CONFIG_USE_AFE_SR
        if (audio_processor_.IsRunning()) {
-            audio_processor_.Input(data, size);
+            audio_processor_.Input(data);
        }
        if (wake_word_detect_.IsDetectionRunning()) {
-            wake_word_detect_.Feed(data, size);
+            wake_word_detect_.Feed(data);
        }
 #else
-        std::vector<int16_t> pcm(data, data + size);
-        Schedule([this, pcm = std::move(pcm)]() {
+        Schedule([this, data = std::move(data)]() {
            if (chat_state_ == kChatStateListening) {
                std::lock_guard<std::mutex> lock(mutex_);
-                audio_encode_queue_.emplace_back(std::move(pcm));
+                audio_encode_queue_.emplace_back(std::move(data));
                cv_.notify_all();
            }
        });
 #endif
    });

-    // Initialize the audio device
-    audio_device_.Start(CONFIG_AUDIO_INPUT_SAMPLE_RATE, CONFIG_AUDIO_OUTPUT_SAMPLE_RATE);
-
    // OPUS encoder / decoder use a lot of stack memory
    const size_t opus_stack_size = 4096 * 8;
-    audio_encode_task_stack_ = (StackType_t*)malloc(opus_stack_size);
+    audio_encode_task_stack_ = (StackType_t*)heap_caps_malloc(opus_stack_size, MALLOC_CAP_SPIRAM);
    audio_encode_task_ = xTaskCreateStatic([](void* arg) {
        Application* app = (Application*)arg;
        app->AudioEncodeTask();
@@ -221,9 +192,92 @@ void Application::Start() {
        Application* app = (Application*)arg;
        app->AudioPlayTask();
        vTaskDelete(NULL);
-    }, "play_audio", 4096 * 4, this, 5, NULL);
+    }, "play_audio", 4096 * 4, this, 4, NULL);
+
+    board.StartNetwork();
+    // Blink the LED to indicate the device is running
+    builtin_led.SetGreen();
+    builtin_led.BlinkOnce();
+
+    boot_button_.OnClick([this]() {
+        Schedule([this]() {
+            if (chat_state_ == kChatStateIdle) {
+                SetChatState(kChatStateConnecting);
+                StartWebSocketClient();
+
+                if (ws_client_ && ws_client_->IsConnected()) {
+                    opus_encoder_.ResetState();
+#ifdef CONFIG_USE_AFE_SR
+                    audio_processor_.Start();
+#endif
+                    SetChatState(kChatStateListening);
+                    ESP_LOGI(TAG, "Communication started");
+                } else {
+                    SetChatState(kChatStateIdle);
+                }
+            } else if (chat_state_ == kChatStateSpeaking) {
+                AbortSpeaking();
+            } else if (chat_state_ == kChatStateListening) {
+                if (ws_client_ && ws_client_->IsConnected()) {
+                    ws_client_->Close();
+                }
+            }
+        });
+    });
+
+    volume_up_button_.OnClick([this]() {
+        Schedule([this]() {
+            auto volume = audio_device_->output_volume() + 10;
+            if (volume > 100) {
+                volume = 100;
+            }
+            audio_device_->SetOutputVolume(volume);
+            display_.ShowNotification("Volume\n" + std::to_string(volume));
+        });
+    });
+
+    volume_up_button_.OnLongPress([this]() {
+        Schedule([this]() {
+            audio_device_->SetOutputVolume(100);
+            display_.ShowNotification("Volume\n100");
+        });
+    });
+
+    volume_down_button_.OnClick([this]() {
+        Schedule([this]() {
+            auto volume = audio_device_->output_volume() - 10;
+            if (volume < 0) {
+                volume = 0;
+            }
+            audio_device_->SetOutputVolume(volume);
+            display_.ShowNotification("Volume\n" + std::to_string(volume));
+        });
+    });
+
+    volume_down_button_.OnLongPress([this]() {
+        Schedule([this]() {
+            audio_device_->SetOutputVolume(0);
+            display_.ShowNotification("Volume\n0");
+        });
+    });
+
+    const size_t main_loop_stack_size = 4096 * 2;
+    main_loop_task_stack_ = (StackType_t*)heap_caps_malloc(main_loop_stack_size, MALLOC_CAP_SPIRAM);
+    xTaskCreateStatic([](void* arg) {
+        Application* app = (Application*)arg;
+        app->MainLoop();
+        vTaskDelete(NULL);
+    }, "main_loop", main_loop_stack_size, this, 1, main_loop_task_stack_, &main_loop_task_buffer_);
+
+    // Launch a task to check for new firmware version
+    xTaskCreate([](void* arg) {
+        Application* app = (Application*)arg;
+        app->CheckNewVersion();
+        vTaskDelete(NULL);
+    }, "check_new_version", 4096 * 2, this, 1, NULL);

 #ifdef CONFIG_USE_AFE_SR
+    wake_word_detect_.Initialize(audio_device_->input_channels(), audio_device_->input_reference());
    wake_word_detect_.OnVadStateChange([this](bool speaking) {
        Schedule([this, speaking]() {
            auto& builtin_led = BuiltinLed::GetInstance();
@@ -263,7 +317,7 @@ void Application::Start() {
                    SetChatState(kChatStateIdle);
                }
            } else if (chat_state_ == kChatStateSpeaking) {
-                break_speaking_ = true;
+                AbortSpeaking();
            }

            // Resume detection
@@ -272,6 +326,7 @@ void Application::Start() {
    });
    wake_word_detect_.StartDetection();

+    audio_processor_.Initialize(audio_device_->input_channels(), audio_device_->input_reference());
    audio_processor_.OnOutput([this](std::vector<int16_t>&& data) {
        Schedule([this, data = std::move(data)]() {
            if (chat_state_ == kChatStateListening) {
@@ -283,79 +338,8 @@ void Application::Start() {
    });
 #endif

-    // Blink the LED to indicate the device is running
-    builtin_led.SetGreen();
-    builtin_led.BlinkOnce();
-
-    boot_button_.OnClick([this]() {
-        Schedule([this]() {
-            if (chat_state_ == kChatStateIdle) {
-                SetChatState(kChatStateConnecting);
-                StartWebSocketClient();
-
-                if (ws_client_ && ws_client_->IsConnected()) {
-                    opus_encoder_.ResetState();
-#ifdef CONFIG_USE_AFE_SR
-                    audio_processor_.Start();
-#endif
-                    SetChatState(kChatStateListening);
-                    ESP_LOGI(TAG, "Communication started");
-                } else {
-                    SetChatState(kChatStateIdle);
-                }
-            } else if (chat_state_ == kChatStateSpeaking) {
-                break_speaking_ = true;
-            } else if (chat_state_ == kChatStateListening) {
-                if (ws_client_ && ws_client_->IsConnected()) {
-                    ws_client_->Close();
-                }
-            }
-        });
-    });
-
-    volume_up_button_.OnClick([this]() {
-        Schedule([this]() {
-            auto volume = audio_device_.output_volume() + 10;
-            if (volume > 100) {
-                volume = 0;
-            }
-            audio_device_.SetOutputVolume(volume);
-#ifdef CONFIG_USE_DISPLAY
-            display_.ShowNotification("Volume\n" + std::to_string(volume));
-#endif
-        });
-    });
-
-    volume_up_button_.OnLongPress([this]() {
-        Schedule([this]() {
-            audio_device_.SetOutputVolume(0);
-#ifdef CONFIG_USE_DISPLAY
-            display_.ShowNotification("Volume\n0");
-#endif
-        });
-    });
-
-    xTaskCreate([](void* arg) {
-        Application* app = (Application*)arg;
-        app->MainLoop();
-        vTaskDelete(NULL);
-    }, "main_loop", 4096 * 2, this, 5, NULL);
-
-    // Launch a task to check for new firmware version
-    xTaskCreate([](void* arg) {
-        Application* app = (Application*)arg;
-        app->CheckNewVersion();
-        vTaskDelete(NULL);
-    }, "check_new_version", 4096 * 2, this, 1, NULL);
-
-#ifdef CONFIG_USE_DISPLAY
-    // Launch a task to update the display
-    xTaskCreate([](void* arg) {
-        Application* app = (Application*)arg;
-        app->UpdateDisplay();
-        vTaskDelete(NULL);
-    }, "update_display", 4096, this, 1, NULL);
-#endif
+    SetChatState(kChatStateIdle);
+    display_.UpdateDisplay();
 }

 void Application::Schedule(std::function<void()> callback) {
@@ -380,8 +364,25 @@ void Application::MainLoop() {
    }
 }

+void Application::AbortSpeaking() {
+    ESP_LOGI(TAG, "Abort speaking");
+    skip_to_end_ = true;
+
+    if (ws_client_ && ws_client_->IsConnected()) {
+        cJSON* root = cJSON_CreateObject();
+        cJSON_AddStringToObject(root, "type", "abort");
+        char* json = cJSON_PrintUnformatted(root);
+
+        std::lock_guard<std::mutex> lock(mutex_);
+        ws_client_->Send(json);
+        cJSON_Delete(root);
+        free(json);
+    }
+}
+
 void Application::SetChatState(ChatState state) {
    const char* state_str[] = {
+        "unknown",
        "idle",
        "connecting",
        "listening",
@@ -389,15 +390,21 @@ void Application::SetChatState(ChatState state) {
        "wake_word_detected",
        "testing",
        "upgrading",
-        "unknown"
+        "invalid_state"
    };
+    if (chat_state_ == state) {
+        // No need to update the state
+        return;
+    }
    chat_state_ = state;
    ESP_LOGI(TAG, "STATE: %s", state_str[chat_state_]);

    auto& builtin_led = BuiltinLed::GetInstance();
    switch (chat_state_) {
+        case kChatStateUnknown:
        case kChatStateIdle:
            builtin_led.TurnOff();
+            audio_device_->EnableOutput(false);
            break;
        case kChatStateConnecting:
            builtin_led.SetBlue();
@@ -410,6 +417,7 @@ void Application::SetChatState(ChatState state) {
        case kChatStateSpeaking:
            builtin_led.SetGreen();
            builtin_led.TurnOn();
+            audio_device_->EnableOutput(true);
            break;
        case kChatStateWakeWordDetected:
            builtin_led.SetBlue();
@@ -434,25 +442,24 @@ void Application::SetChatState(ChatState state) {
    }
 }

-BinaryProtocol* Application::AllocateBinaryProtocol(const uint8_t* payload, size_t payload_size) {
-    auto last_timestamp = 0;
-    auto protocol = (BinaryProtocol*)heap_caps_malloc(sizeof(BinaryProtocol) + payload_size, MALLOC_CAP_SPIRAM);
-    protocol->version = htons(PROTOCOL_VERSION);
-    protocol->type = htons(0);
+BinaryProtocol3* Application::AllocateBinaryProtocol3(const uint8_t* payload, size_t payload_size) {
+    auto protocol = (BinaryProtocol3*)heap_caps_malloc(sizeof(BinaryProtocol3) + payload_size, MALLOC_CAP_SPIRAM);
+    protocol->type = 0;
    protocol->reserved = 0;
-    protocol->timestamp = htonl(last_timestamp);
-    protocol->payload_size = htonl(payload_size);
-    assert(sizeof(BinaryProtocol) == 16);
+    protocol->payload_size = htons(payload_size);
+    assert(sizeof(BinaryProtocol3) == 4UL);
    memcpy(protocol->payload, payload, payload_size);
    return protocol;
 }

 void Application::AudioEncodeTask() {
    ESP_LOGI(TAG, "Audio encode task started");
+    const int max_audio_play_queue_size_ = 2;
+
    while (true) {
        std::unique_lock<std::mutex> lock(mutex_);
        cv_.wait(lock, [this]() {
-            return !audio_encode_queue_.empty() || !audio_decode_queue_.empty();
+            return !audio_encode_queue_.empty() || (!audio_decode_queue_.empty() && audio_play_queue_.size() < max_audio_play_queue_size_);
        });

        if (!audio_encode_queue_.empty()) {
@@ -462,10 +469,10 @@ void Application::AudioEncodeTask() {

            // Encode audio data
            opus_encoder_.Encode(pcm, [this](const uint8_t* opus, size_t opus_size) {
-                auto protocol = AllocateBinaryProtocol(opus, opus_size);
+                auto protocol = AllocateBinaryProtocol3(opus, opus_size);
                Schedule([this, protocol, opus_size]() {
                    if (ws_client_ && ws_client_->IsConnected()) {
-                        if (!ws_client_->Send(protocol, sizeof(BinaryProtocol) + opus_size, true)) {
+                        if (!ws_client_->Send(protocol, sizeof(BinaryProtocol3) + opus_size, true)) {
                            ESP_LOGE(TAG, "Failed to send audio data");
                        }
                    }
@@ -477,21 +484,23 @@ void Application::AudioEncodeTask() {
            audio_decode_queue_.pop_front();
            lock.unlock();

-            int frame_size = opus_decode_sample_rate_ * opus_duration_ms_ / 1000;
-            packet->pcm.resize(frame_size);
+            if (packet->type == kAudioPacketTypeData && !skip_to_end_) {
+                int frame_size = opus_decode_sample_rate_ * opus_duration_ms_ / 1000;
+                packet->pcm.resize(frame_size);

-            int ret = opus_decode(opus_decoder_, packet->opus.data(), packet->opus.size(), packet->pcm.data(), frame_size, 0);
-            if (ret < 0) {
-                ESP_LOGE(TAG, "Failed to decode audio, error code: %d", ret);
-                delete packet;
-                continue;
-            }
+                int ret = opus_decode(opus_decoder_, packet->opus.data(), packet->opus.size(), packet->pcm.data(), frame_size, 0);
+                if (ret < 0) {
+                    ESP_LOGE(TAG, "Failed to decode audio, error code: %d", ret);
+                    delete packet;
+                    continue;
+                }

-            if (opus_decode_sample_rate_ != CONFIG_AUDIO_OUTPUT_SAMPLE_RATE) {
-                int target_size = opus_resampler_.GetOutputSamples(frame_size);
-                std::vector<int16_t> resampled(target_size);
-                opus_resampler_.Process(packet->pcm.data(), frame_size, resampled.data());
-                packet->pcm = std::move(resampled);
+                if (opus_decode_sample_rate_ != AUDIO_OUTPUT_SAMPLE_RATE) {
+                    int target_size = output_resampler_.GetOutputSamples(frame_size);
+                    std::vector<int16_t> resampled(target_size);
+                    output_resampler_.Process(packet->pcm.data(), frame_size, resampled.data());
+                    packet->pcm = std::move(resampled);
+                }
            }

            std::lock_guard<std::mutex> lock(mutex_);
@@ -510,35 +519,32 @@ void Application::HandleAudioPacket(AudioPacket* packet) {
        }

        // This will block until the audio device has finished playing the audio
-        audio_device_.OutputData(packet->pcm);
-
-        if (break_speaking_) {
-            break_speaking_ = false;
-            skip_to_end_ = true;
-            
-            // Play a silence and skip to the end
-            int frame_size = opus_decode_sample_rate_ / 1000 * opus_duration_ms_;
-            std::vector<int16_t> silence(frame_size);
-            bzero(silence.data(), silence.size() * sizeof(int16_t));
-            audio_device_.OutputData(silence);
-        }
+        audio_device_->OutputData(packet->pcm);
        break;
    }
    case kAudioPacketTypeStart:
+        break_speaking_ = false;
+        skip_to_end_ = false;
        Schedule([this]() {
            SetChatState(kChatStateSpeaking);
        });
        break;
    case kAudioPacketTypeStop:
-        skip_to_end_ = false;
        Schedule([this]() {
-            SetChatState(kChatStateListening);
+            if (ws_client_ && ws_client_->IsConnected()) {
+                SetChatState(kChatStateListening);
+            } else {
+                SetChatState(kChatStateIdle);
+            }
        });
        break;
    case kAudioPacketTypeSentenceStart:
        ESP_LOGI(TAG, "<< %s", packet->text.c_str());
        break;
    case kAudioPacketTypeSentenceEnd:
+        if (break_speaking_) {
+            skip_to_end_ = true;
+        }
        break;
    default:
        ESP_LOGI(TAG, "Unknown packet type: %d", packet->type);
@@ -558,6 +564,7 @@ void Application::AudioPlayTask() {
        });
        auto packet = std::move(audio_play_queue_.front());
        audio_play_queue_.pop_front();
+        cv_.notify_all();
        lock.unlock();

        HandleAudioPacket(packet);
@@ -572,9 +579,26 @@ void Application::SetDecodeSampleRate(int sample_rate) {
    opus_decoder_destroy(opus_decoder_);
    opus_decode_sample_rate_ = sample_rate;
    opus_decoder_ = opus_decoder_create(opus_decode_sample_rate_, 1, NULL);
-    if (opus_decode_sample_rate_ != CONFIG_AUDIO_OUTPUT_SAMPLE_RATE) {
-        ESP_LOGI(TAG, "Resampling audio from %d to %d", opus_decode_sample_rate_, CONFIG_AUDIO_OUTPUT_SAMPLE_RATE);
-        opus_resampler_.Configure(opus_decode_sample_rate_, CONFIG_AUDIO_OUTPUT_SAMPLE_RATE);
+    if (opus_decode_sample_rate_ != AUDIO_OUTPUT_SAMPLE_RATE) {
+        ESP_LOGI(TAG, "Resampling audio from %d to %d", opus_decode_sample_rate_, AUDIO_OUTPUT_SAMPLE_RATE);
+        output_resampler_.Configure(opus_decode_sample_rate_, AUDIO_OUTPUT_SAMPLE_RATE);
+    }
+}
+
+void Application::ParseBinaryProtocol3(const char* data, size_t size) {
+    for (const char* p = data; p < data + size; ) {
+        auto protocol = (BinaryProtocol3*)p;
+        p += sizeof(BinaryProtocol3);
+
+        auto packet = new AudioPacket();
+        packet->type = kAudioPacketTypeData;
+        auto payload_size = ntohs(protocol->payload_size);
+        packet->opus.resize(payload_size);
+        memcpy(packet->opus.data(), protocol->payload, payload_size);
+        p += payload_size;
+
+        std::lock_guard<std::mutex> lock(mutex_);
+        audio_decode_queue_.push_back(packet);
    }
 }

@@ -586,18 +610,10 @@ void Application::StartWebSocketClient() {

    std::string url = CONFIG_WEBSOCKET_URL;
    std::string token = "Bearer " + std::string(CONFIG_WEBSOCKET_ACCESS_TOKEN);
-#ifdef CONFIG_USE_ML307
-    ws_client_ = new WebSocket(new Ml307SslTransport(ml307_at_modem_, 0));
-#else
-    if (url.find("wss://") == 0) {
-        ws_client_ = new WebSocket(new TlsTransport());
-    } else {
-        ws_client_ = new WebSocket(new TcpTransport());
-    }
-#endif
+    ws_client_ = Board::GetInstance().CreateWebSocket();
    ws_client_->SetHeader("Authorization", token.c_str());
-    ws_client_->SetHeader("Device-Id", SystemInfo::GetMacAddress().c_str());
    ws_client_->SetHeader("Protocol-Version", std::to_string(PROTOCOL_VERSION).c_str());
+    ws_client_->SetHeader("Device-Id", SystemInfo::GetMacAddress().c_str());

    ws_client_->OnConnected([this]() {
        ESP_LOGI(TAG, "Websocket connected");
@@ -607,24 +623,14 @@ void Application::StartWebSocketClient() {
        std::string message = "{";
        message += "\"type\":\"hello\",";
        message += "\"audio_params\":{";
-        message += "\"format\":\"opus\", \"sample_rate\":" + std::to_string(CONFIG_AUDIO_INPUT_SAMPLE_RATE) + ", \"channels\":1";
+        message += "\"format\":\"opus\", \"sample_rate\":16000, \"channels\":1";
        message += "}}";
        ws_client_->Send(message);
    });

    ws_client_->OnData([this](const char* data, size_t len, bool binary) {
        if (binary) {
-            auto protocol = (BinaryProtocol*)data;
-
-            auto packet = new AudioPacket();
-            packet->type = kAudioPacketTypeData;
-            packet->timestamp = ntohl(protocol->timestamp);
-            auto payload_size = ntohl(protocol->payload_size);
-            packet->opus.resize(payload_size);
-            memcpy(packet->opus.data(), protocol->payload, payload_size);
-
-            std::lock_guard<std::mutex> lock(mutex_);
-            audio_decode_queue_.push_back(packet);
+            ParseBinaryProtocol3(data, len);
            cv_.notify_all();
        } else {
            // Parse JSON data
@@ -640,6 +646,9 @@ void Application::StartWebSocketClient() {
                        if (sample_rate != NULL) {
                            SetDecodeSampleRate(sample_rate->valueint);
                        }
+
+                        // If the device is speaking, we need to skip the last session
+                        skip_to_end_ = true;
                    } else if (strcmp(state->valuestring, "stop") == 0) {
                        packet->type = kAudioPacketTypeStop;
                    } else if (strcmp(state->valuestring, "sentence_end") == 0) {
--- a/main/Application.h
+++ b/main/Application.h
@@ -1,13 +1,9 @@
 #ifndef _APPLICATION_H_
 #define _APPLICATION_H_

-#include "AudioDevice.h"
 #include <OpusEncoder.h>
 #include <OpusResampler.h>
 #include <WebSocket.h>
-#include <Ml307AtModem.h>
-#include <Ml307Http.h>
-#include <EspHttp.h>

 #include <opus.h>
 #include <resampler_structs.h>
@@ -17,7 +13,9 @@
 #include <list>
 #include <condition_variable>

+#include "AudioDevice.h"
 #include "Display.h"
+#include "Board.h"
 #include "FirmwareUpgrade.h"

 #ifdef CONFIG_USE_AFE_SR
@@ -30,13 +28,11 @@
 #define DETECTION_RUNNING 1
 #define COMMUNICATION_RUNNING 2

-#define PROTOCOL_VERSION 2
-struct BinaryProtocol {
-    uint16_t version;
-    uint16_t type;
-    uint32_t reserved;
-    uint32_t timestamp;
-    uint32_t payload_size;
+#define PROTOCOL_VERSION 3
+struct BinaryProtocol3 {
+    uint8_t type;
+    uint8_t reserved;
+    uint16_t payload_size;
    uint8_t payload[];
 } __attribute__((packed));

@@ -59,6 +55,7 @@ struct AudioPacket {


 enum ChatState {
+    kChatStateUnknown,
    kChatStateIdle,
    kChatStateConnecting,
    kChatStateListening,
@@ -75,7 +72,12 @@ public:
    }

    void Start();
-
+    ChatState GetChatState() const { return chat_state_; }
+    Display& GetDisplay() { return display_; }
+    void Schedule(std::function<void()> callback);
+    void SetChatState(ChatState state);
+    void Alert(const std::string&& title, const std::string&& message);
+    void AbortSpeaking();
    // 删除拷贝构造函数和赋值运算符
    Application(const Application&) = delete;
    Application& operator=(const Application&) = delete;
@@ -86,29 +88,23 @@ private:

    Button boot_button_;
    Button volume_up_button_;
-    AudioDevice audio_device_;
+    Button volume_down_button_;
+    AudioDevice* audio_device_ = nullptr;
+    Display display_;
 #ifdef CONFIG_USE_AFE_SR
    WakeWordDetect wake_word_detect_;
    AudioProcessor audio_processor_;
-#endif
-#ifdef CONFIG_USE_ML307
-    Ml307AtModem ml307_at_modem_;
-    Ml307Http http_;
-#else
-    EspHttp http_;
 #endif
    FirmwareUpgrade firmware_upgrade_;
-#ifdef CONFIG_USE_DISPLAY
-    Display display_;
-#endif
    std::mutex mutex_;
    std::condition_variable_any cv_;
    std::list<std::function<void()>> main_tasks_;
    WebSocket* ws_client_ = nullptr;
    EventGroupHandle_t event_group_;
-    volatile ChatState chat_state_ = kChatStateIdle;
+    volatile ChatState chat_state_ = kChatStateUnknown;
    volatile bool break_speaking_ = false;
    bool skip_to_end_ = false;
+    esp_timer_handle_t update_display_timer_ = nullptr;

    // Audio encode / decode
    TaskHandle_t audio_encode_task_ = nullptr;
@@ -122,25 +118,25 @@ private:
    OpusDecoder* opus_decoder_ = nullptr;

    int opus_duration_ms_ = 60;
-    int opus_decode_sample_rate_ = CONFIG_AUDIO_OUTPUT_SAMPLE_RATE;
-    OpusResampler opus_resampler_;
+    int opus_decode_sample_rate_ = AUDIO_OUTPUT_SAMPLE_RATE;
+    OpusResampler input_resampler_;
+    OpusResampler output_resampler_;

-    TaskHandle_t check_new_version_task_ = nullptr;
-    StaticTask_t check_new_version_task_buffer_;
-    StackType_t* check_new_version_task_stack_ = nullptr;
+    TaskHandle_t main_loop_task_ = nullptr;
+    StaticTask_t main_loop_task_buffer_;
+    StackType_t* main_loop_task_stack_ = nullptr;

    void MainLoop();
-    void Schedule(std::function<void()> callback);
-    BinaryProtocol* AllocateBinaryProtocol(const uint8_t* payload, size_t payload_size);
+    BinaryProtocol3* AllocateBinaryProtocol3(const uint8_t* payload, size_t payload_size);
+    void ParseBinaryProtocol3(const char* data, size_t size);
    void SetDecodeSampleRate(int sample_rate);
-    void SetChatState(ChatState state);
    void StartWebSocketClient();
    void CheckNewVersion();
-    void UpdateDisplay();

    void AudioEncodeTask();
    void AudioPlayTask();
    void HandleAudioPacket(AudioPacket* packet);
+    void PlayLocalFile(const char* data, size_t size);
 };

 #endif // _APPLICATION_H_
--- a/main/AudioDevice.cc
+++ b/main/AudioDevice.cc
@@ -1,10 +1,15 @@
 #include "AudioDevice.h"
+#include "Board.h"
+
 #include <esp_log.h>
 #include <cstring>
 #include <cmath>
+
 #define TAG "AudioDevice"

-AudioDevice::AudioDevice() {
+AudioDevice::AudioDevice()
+    : input_sample_rate_(AUDIO_INPUT_SAMPLE_RATE),
+      output_sample_rate_(AUDIO_OUTPUT_SAMPLE_RATE) {
 }

 AudioDevice::~AudioDevice() {
@@ -19,26 +24,16 @@ AudioDevice::~AudioDevice() {
    }
 }

-void AudioDevice::Start(int input_sample_rate, int output_sample_rate) {
-    input_sample_rate_ = input_sample_rate;
-    output_sample_rate_ = output_sample_rate;
-
-#ifdef CONFIG_AUDIO_DEVICE_I2S_SIMPLEX
-        CreateSimplexChannels();
+void AudioDevice::Initialize() {
+#ifdef AUDIO_I2S_METHOD_SIMPLEX
+    CreateSimplexChannels();
 #else
-        CreateDuplexChannels();
+    CreateDuplexChannels();
 #endif
-
-    ESP_ERROR_CHECK(i2s_channel_enable(tx_handle_));
-    ESP_ERROR_CHECK(i2s_channel_enable(rx_handle_));
-
-    xTaskCreate([](void* arg) {
-        auto audio_device = (AudioDevice*)arg;
-        audio_device->InputTask();
-    }, "audio_input", 4096 * 2, this, 5, &audio_input_task_);
 }

 void AudioDevice::CreateDuplexChannels() {
+#ifndef AUDIO_I2S_METHOD_SIMPLEX
    duplex_ = true;

    i2s_chan_config_t chan_cfg = {
@@ -73,10 +68,10 @@ void AudioDevice::CreateDuplexChannels() {
        },
        .gpio_cfg = {
            .mclk = I2S_GPIO_UNUSED,
-            .bclk = (gpio_num_t)CONFIG_AUDIO_DEVICE_I2S_MIC_GPIO_BCLK,
-            .ws = (gpio_num_t)CONFIG_AUDIO_DEVICE_I2S_MIC_GPIO_WS,
-            .dout = (gpio_num_t)CONFIG_AUDIO_DEVICE_I2S_SPK_GPIO_DOUT,
-            .din = (gpio_num_t)CONFIG_AUDIO_DEVICE_I2S_MIC_GPIO_DIN,
+            .bclk = (gpio_num_t)AUDIO_I2S_GPIO_BCLK,
+            .ws = (gpio_num_t)AUDIO_I2S_GPIO_LRCK,
+            .dout = (gpio_num_t)AUDIO_I2S_GPIO_DOUT,
+            .din = (gpio_num_t)AUDIO_I2S_GPIO_DIN,
            .invert_flags = {
                .mclk_inv = false,
                .bclk_inv = false,
@@ -86,18 +81,21 @@ void AudioDevice::CreateDuplexChannels() {
    };
    ESP_ERROR_CHECK(i2s_channel_init_std_mode(tx_handle_, &std_cfg));
    ESP_ERROR_CHECK(i2s_channel_init_std_mode(rx_handle_, &std_cfg));
+    ESP_ERROR_CHECK(i2s_channel_enable(tx_handle_));
+    ESP_ERROR_CHECK(i2s_channel_enable(rx_handle_));
    ESP_LOGI(TAG, "Duplex channels created");
+#endif
 }

-#ifdef CONFIG_AUDIO_DEVICE_I2S_SIMPLEX
 void AudioDevice::CreateSimplexChannels() {
+#ifdef AUDIO_I2S_METHOD_SIMPLEX
    // Create a new channel for speaker
    i2s_chan_config_t chan_cfg = {
        .id = I2S_NUM_0,
        .role = I2S_ROLE_MASTER,
        .dma_desc_num = 6,
        .dma_frame_num = 240,
-        .auto_clear_after_cb = false,
+        .auto_clear_after_cb = true,
        .auto_clear_before_cb = false,
        .intr_priority = 0,
    };
@@ -124,9 +122,9 @@ void AudioDevice::CreateSimplexChannels() {
        },
        .gpio_cfg = {
            .mclk = I2S_GPIO_UNUSED,
-            .bclk = (gpio_num_t)CONFIG_AUDIO_DEVICE_I2S_SPK_GPIO_BCLK,
-            .ws = (gpio_num_t)CONFIG_AUDIO_DEVICE_I2S_SPK_GPIO_WS,
-            .dout = (gpio_num_t)CONFIG_AUDIO_DEVICE_I2S_SPK_GPIO_DOUT,
+            .bclk = (gpio_num_t)AUDIO_I2S_SPK_GPIO_BCLK,
+            .ws = (gpio_num_t)AUDIO_I2S_SPK_GPIO_LRCK,
+            .dout = (gpio_num_t)AUDIO_I2S_SPK_GPIO_DOUT,
            .din = I2S_GPIO_UNUSED,
            .invert_flags = {
                .mclk_inv = false,
@@ -141,48 +139,67 @@ void AudioDevice::CreateSimplexChannels() {
    chan_cfg.id = I2S_NUM_1;
    ESP_ERROR_CHECK(i2s_new_channel(&chan_cfg, nullptr, &rx_handle_));
    std_cfg.clk_cfg.sample_rate_hz = (uint32_t)input_sample_rate_;
-    std_cfg.gpio_cfg.bclk = (gpio_num_t)CONFIG_AUDIO_DEVICE_I2S_MIC_GPIO_BCLK;
-    std_cfg.gpio_cfg.ws = (gpio_num_t)CONFIG_AUDIO_DEVICE_I2S_MIC_GPIO_WS;
+    std_cfg.gpio_cfg.bclk = (gpio_num_t)AUDIO_I2S_MIC_GPIO_SCK;
+    std_cfg.gpio_cfg.ws = (gpio_num_t)AUDIO_I2S_MIC_GPIO_WS;
    std_cfg.gpio_cfg.dout = I2S_GPIO_UNUSED;
-    std_cfg.gpio_cfg.din = (gpio_num_t)CONFIG_AUDIO_DEVICE_I2S_MIC_GPIO_DIN;
+    std_cfg.gpio_cfg.din = (gpio_num_t)AUDIO_I2S_MIC_GPIO_DIN;
    ESP_ERROR_CHECK(i2s_channel_init_std_mode(rx_handle_, &std_cfg));
-    ESP_LOGI(TAG, "Simplex channels created");
-}
-#endif

-void AudioDevice::Write(const int16_t* data, int samples) {
+    ESP_ERROR_CHECK(i2s_channel_enable(tx_handle_));
+    ESP_ERROR_CHECK(i2s_channel_enable(rx_handle_));
+    ESP_LOGI(TAG, "Simplex channels created");
+#endif
+}
+
+int AudioDevice::Write(const int16_t* data, int samples) {
    int32_t buffer[samples];

    // output_volume_: 0-100
    // volume_factor_: 0-65536
    int32_t volume_factor = pow(double(output_volume_) / 100.0, 2) * 65536;
    for (int i = 0; i < samples; i++) {
-        buffer[i] = int32_t(data[i]) * volume_factor;
+        int64_t temp = int64_t(data[i]) * volume_factor; // 使用 int64_t 进行乘法运算
+        if (temp > INT32_MAX) {
+            buffer[i] = INT32_MAX;
+        } else if (temp < INT32_MIN) {
+            buffer[i] = INT32_MIN;
+        } else {
+            buffer[i] = static_cast<int32_t>(temp);
+        }
    }

    size_t bytes_written;
    ESP_ERROR_CHECK(i2s_channel_write(tx_handle_, buffer, samples * sizeof(int32_t), &bytes_written, portMAX_DELAY));
+    return bytes_written / sizeof(int32_t);
 }

 int AudioDevice::Read(int16_t* dest, int samples) {
    size_t bytes_read;

-    int32_t bit32_buffer_[samples];
-    if (i2s_channel_read(rx_handle_, bit32_buffer_, samples * sizeof(int32_t), &bytes_read, portMAX_DELAY) != ESP_OK) {
+    int32_t bit32_buffer[samples];
+    if (i2s_channel_read(rx_handle_, bit32_buffer, samples * sizeof(int32_t), &bytes_read, portMAX_DELAY) != ESP_OK) {
        ESP_LOGE(TAG, "Read Failed!");
        return 0;
    }

    samples = bytes_read / sizeof(int32_t);
    for (int i = 0; i < samples; i++) {
-        int32_t value = bit32_buffer_[i] >> 12;
+        int32_t value = bit32_buffer[i] >> 12;
        dest[i] = (value > INT16_MAX) ? INT16_MAX : (value < -INT16_MAX) ? -INT16_MAX : (int16_t)value;
    }
    return samples;
 }

-void AudioDevice::OnInputData(std::function<void(const int16_t*, int)> callback) {
+void AudioDevice::OnInputData(std::function<void(std::vector<int16_t>&& data)> callback) {
    on_input_data_ = callback;
+
+    // 创建音频输入任务
+    if (audio_input_task_ == nullptr) {
+        xTaskCreate([](void* arg) {
+            auto audio_device = (AudioDevice*)arg;
+            audio_device->InputTask();
+        }, "audio_input", 4096 * 2, this, 3, &audio_input_task_);
+    }
 }

 void AudioDevice::OutputData(std::vector<int16_t>& data) {
@@ -191,12 +208,14 @@ void AudioDevice::OutputData(std::vector<int16_t>& data) {

 void AudioDevice::InputTask() {
    int duration = 30;
-    int input_frame_size = input_sample_rate_ / 1000 * duration;
-    int16_t input_buffer[input_frame_size];
+    int input_frame_size = input_sample_rate_ / 1000 * duration * input_channels_;
    while (true) {
-        int samples = Read(input_buffer, input_frame_size);
+        std::vector<int16_t> input_data(input_frame_size);
+        int samples = Read(input_data.data(), input_data.size());
        if (samples > 0) {
-            on_input_data_(input_buffer, samples);
+            if (on_input_data_) {
+                on_input_data_(std::move(input_data));
+            }
        }
    }
 }
@@ -205,3 +224,19 @@ void AudioDevice::SetOutputVolume(int volume) {
    output_volume_ = volume;
    ESP_LOGI(TAG, "Set output volume to %d", output_volume_);
 }
+
+void AudioDevice::EnableInput(bool enable) {
+    if (enable == input_enabled_) {
+        return;
+    }
+    input_enabled_ = enable;
+    ESP_LOGI(TAG, "Set input enable to %s", enable ? "true" : "false");
+}
+
+void AudioDevice::EnableOutput(bool enable) {
+    if (enable == output_enabled_) {
+        return;
+    }
+    output_enabled_ = enable;
+    ESP_LOGI(TAG, "Set output enable to %s", enable ? "true" : "false");
+}
--- a/main/AudioDevice.h
+++ b/main/AudioDevice.h
@@ -2,7 +2,6 @@
 #define _AUDIO_DEVICE_H

 #include <freertos/FreeRTOS.h>
-#include <freertos/event_groups.h>
 #include <driver/i2s_std.h>

 #include <vector>
@@ -12,35 +11,46 @@
 class AudioDevice {
 public:
    AudioDevice();
-    ~AudioDevice();
+    virtual ~AudioDevice();
+    virtual void Initialize();

-    void Start(int input_sample_rate, int output_sample_rate);
-    void OnInputData(std::function<void(const int16_t*, int)> callback);
+    void OnInputData(std::function<void(std::vector<int16_t>&& data)> callback);
    void OutputData(std::vector<int16_t>& data);
-    void SetOutputVolume(int volume);
+    virtual void SetOutputVolume(int volume);
+    virtual void EnableInput(bool enable);
+    virtual void EnableOutput(bool enable);
+
+    inline bool duplex() const { return duplex_; }
+    inline bool input_reference() const { return input_reference_; }
+    inline int input_sample_rate() const { return input_sample_rate_; }
+    inline int output_sample_rate() const { return output_sample_rate_; }
+    inline int input_channels() const { return input_channels_; }
+    inline int output_channels() const { return output_channels_; }
+    inline int output_volume() const { return output_volume_; }

-    int input_sample_rate() const { return input_sample_rate_; }
-    int output_sample_rate() const { return output_sample_rate_; }
-    bool duplex() const { return duplex_; }
-    int output_volume() const { return output_volume_; }
 private:
+    TaskHandle_t audio_input_task_ = nullptr;
+    std::function<void(std::vector<int16_t>&& data)> on_input_data_; 
+
+    void InputTask();
+    void CreateSimplexChannels();
+
+protected:
    bool duplex_ = false;
+    bool input_reference_ = false;
+    bool input_enabled_ = false;
+    bool output_enabled_ = false;
    int input_sample_rate_ = 0;
    int output_sample_rate_ = 0;
-    int output_volume_ = 80;
+    int input_channels_ = 1;
+    int output_channels_ = 1;
+    int output_volume_ = 70;
    i2s_chan_handle_t tx_handle_ = nullptr;
    i2s_chan_handle_t rx_handle_ = nullptr;

-    TaskHandle_t audio_input_task_ = nullptr;
-    
-    EventGroupHandle_t event_group_;
-    std::function<void(const int16_t*, int)> on_input_data_;
-
-    void CreateDuplexChannels();
-    void CreateSimplexChannels();
-    void InputTask();
-    int Read(int16_t* dest, int samples);
-    void Write(const int16_t* data, int samples);
+    virtual void CreateDuplexChannels();
+    virtual int Read(int16_t* dest, int samples);
+    virtual int Write(const int16_t* data, int samples);
 };

 #endif // _AUDIO_DEVICE_H
--- a/main/AudioProcessor.cc
+++ b/main/AudioProcessor.cc
@@ -8,6 +8,12 @@ static const char* TAG = "AudioProcessor";
 AudioProcessor::AudioProcessor()
    : afe_communication_data_(nullptr) {
    event_group_ = xEventGroupCreate();
+}
+
+void AudioProcessor::Initialize(int channels, bool reference) {
+    channels_ = channels;
+    reference_ = reference;
+    int ref_num = reference_ ? 1 : 0;

    afe_config_t afe_config = {
        .aec_init = false,
@@ -22,17 +28,17 @@ AudioProcessor::AudioProcessor()
        .wakenet_model_name_2 = NULL,
        .wakenet_mode = DET_MODE_90,
        .afe_mode = SR_MODE_HIGH_PERF,
-        .afe_perferred_core = 0,
-        .afe_perferred_priority = 5,
+        .afe_perferred_core = 1,
+        .afe_perferred_priority = 1,
        .afe_ringbuf_size = 50,
        .memory_alloc_mode = AFE_MEMORY_ALLOC_MORE_PSRAM,
        .afe_linear_gain = 1.0,
        .agc_mode = AFE_MN_PEAK_AGC_MODE_2,
        .pcm_config = {
-            .total_ch_num = 1,
-            .mic_num = 1,
-            .ref_num = 0,
-            .sample_rate = CONFIG_AUDIO_INPUT_SAMPLE_RATE,
+            .total_ch_num = channels_,
+            .mic_num = channels_ - ref_num,
+            .ref_num = ref_num,
+            .sample_rate = 16000,
        },
        .debug_init = false,
        .debug_hook = {{ AFE_DEBUG_HOOK_MASE_TASK_IN, NULL }, { AFE_DEBUG_HOOK_FETCH_TASK_IN, NULL }},
@@ -47,7 +53,7 @@ AudioProcessor::AudioProcessor()
        auto this_ = (AudioProcessor*)arg;
        this_->AudioProcessorTask();
        vTaskDelete(NULL);
-    }, "audio_communication", 4096 * 2, this, 5, NULL);
+    }, "audio_communication", 4096 * 2, this, 1, NULL);
 }

 AudioProcessor::~AudioProcessor() {
@@ -57,10 +63,10 @@ AudioProcessor::~AudioProcessor() {
    vEventGroupDelete(event_group_);
 }

-void AudioProcessor::Input(const int16_t* data, int size) {
-    input_buffer_.insert(input_buffer_.end(), data, data + size);
+void AudioProcessor::Input(std::vector<int16_t>& data) {
+    input_buffer_.insert(input_buffer_.end(), data.begin(), data.end());

-    auto chunk_size = esp_afe_vc_v1.get_feed_chunksize(afe_communication_data_);
+    auto chunk_size = esp_afe_vc_v1.get_feed_chunksize(afe_communication_data_) * channels_;
    while (input_buffer_.size() >= chunk_size) {
        auto chunk = input_buffer_.data();
        esp_afe_vc_v1.feed(afe_communication_data_, chunk);
@@ -92,6 +98,9 @@ void AudioProcessor::AudioProcessorTask() {
        xEventGroupWaitBits(event_group_, PROCESSOR_RUNNING, pdFALSE, pdTRUE, portMAX_DELAY);

        auto res = esp_afe_vc_v1.fetch(afe_communication_data_);
+        if ((xEventGroupGetBits(event_group_) & PROCESSOR_RUNNING) == 0) {
+            continue;
+        }
        if (res == nullptr || res->ret_value == ESP_FAIL) {
            if (res != nullptr) {
                ESP_LOGI(TAG, "Error code: %d", res->ret_value);
--- a/main/AudioProcessor.h
+++ b/main/AudioProcessor.h
@@ -15,7 +15,8 @@ public:
    AudioProcessor();
    ~AudioProcessor();

-    void Input(const int16_t* data, int size);
+    void Initialize(int channels, bool reference);
+    void Input(std::vector<int16_t>& data);
    void Start();
    void Stop();
    bool IsRunning();
@@ -26,6 +27,8 @@ private:
    esp_afe_sr_data_t* afe_communication_data_ = nullptr;
    std::vector<int16_t> input_buffer_;
    std::function<void(std::vector<int16_t>&& data)> output_callback_;
+    int channels_;
+    bool reference_;

    void AudioProcessorTask();
 };
--- a/main/Board.cc
+++ b/main/Board.cc
@@ -0,0 +1,8 @@
+#include "Board.h"
+#include <esp_log.h>
+
+// static const char *TAG = "Board";
+
+bool Board::GetBatteryVoltage(int &voltage, bool& charging) {
+    return false;
+}
--- a/main/Board.h
+++ b/main/Board.h
@@ -0,0 +1,45 @@
+#ifndef BOARD_H
+#define BOARD_H
+
+#include "config.h"
+#include <Http.h>
+#include <WebSocket.h>
+#include <AudioDevice.h>
+#include <string>
+
+void* create_board();
+
+class Board {
+public:
+    static Board& GetInstance() {
+        static Board* instance = nullptr;
+        if (nullptr == instance) {
+            instance = static_cast<Board*>(create_board());
+        }
+        return *instance;
+    }
+
+    virtual void Initialize() = 0;
+    virtual void StartNetwork() = 0;
+    virtual ~Board() = default;
+    virtual AudioDevice* CreateAudioDevice() = 0;
+    virtual Http* CreateHttp() = 0;
+    virtual WebSocket* CreateWebSocket() = 0;
+    virtual bool GetNetworkState(std::string& network_name, int& signal_quality, std::string& signal_quality_text) = 0;
+    virtual bool GetBatteryVoltage(int &voltage, bool& charging);
+    virtual std::string GetJson() = 0;
+
+protected:
+    Board() = default;
+
+private:
+    Board(const Board&) = delete; // 禁用拷贝构造函数
+    Board& operator=(const Board&) = delete; // 禁用赋值操作
+};
+
+#define DECLARE_BOARD(BOARD_CLASS_NAME) \
+void* create_board() { \
+    return new BOARD_CLASS_NAME(); \
+}
+
+#endif // BOARD_H
--- a/main/BoxAudioDevice.cc
+++ b/main/BoxAudioDevice.cc
@@ -0,0 +1,282 @@
+#include "BoxAudioDevice.h"
+#include "Board.h"
+
+#include <esp_log.h>
+#include <cassert>
+
+static const char* TAG = "BoxAudioDevice";
+
+BoxAudioDevice::BoxAudioDevice() {
+}
+
+BoxAudioDevice::~BoxAudioDevice() {
+    ESP_ERROR_CHECK(esp_codec_dev_close(output_dev_));
+    esp_codec_dev_delete(output_dev_);
+    ESP_ERROR_CHECK(esp_codec_dev_close(input_dev_));
+    esp_codec_dev_delete(input_dev_);
+
+    audio_codec_delete_codec_if(in_codec_if_);
+    audio_codec_delete_ctrl_if(in_ctrl_if_);
+    audio_codec_delete_codec_if(out_codec_if_);
+    audio_codec_delete_ctrl_if(out_ctrl_if_);
+    audio_codec_delete_gpio_if(gpio_if_);
+    audio_codec_delete_data_if(data_if_);
+
+    ESP_ERROR_CHECK(i2c_del_master_bus(i2c_master_handle_));
+}
+
+void BoxAudioDevice::Initialize() {
+    duplex_ = true; // 是否双工
+    input_reference_ = AUDIO_INPUT_REFERENCE; // 是否使用参考输入，实现回声消除
+    input_channels_ = input_reference_ ? 2 : 1; // 输入通道数
+
+    // Initialize I2C peripheral
+    i2c_master_bus_config_t i2c_bus_cfg = {
+        .i2c_port = I2C_NUM_1,
+        .sda_io_num = (gpio_num_t)AUDIO_CODEC_I2C_SDA_PIN,
+        .scl_io_num = (gpio_num_t)AUDIO_CODEC_I2C_SCL_PIN,
+        .clk_source = I2C_CLK_SRC_DEFAULT,
+        .glitch_ignore_cnt = 7,
+        .intr_priority = 0,
+        .trans_queue_depth = 0,
+        .flags = {
+            .enable_internal_pullup = 1,
+        },
+    };
+    ESP_ERROR_CHECK(i2c_new_master_bus(&i2c_bus_cfg, &i2c_master_handle_));
+
+    CreateDuplexChannels();
+
+#ifdef AUDIO_CODEC_USE_PCA9557
+    // Initialize PCA9557
+    i2c_device_config_t pca9557_cfg = {
+        .dev_addr_length = I2C_ADDR_BIT_LEN_7,
+        .device_address = 0x19,
+        .scl_speed_hz = 400000,
+        .scl_wait_us = 0,
+        .flags = {
+            .disable_ack_check = 0,
+        },
+    };
+    i2c_master_dev_handle_t pca9557_handle;
+    ESP_ERROR_CHECK(i2c_master_bus_add_device(i2c_master_handle_, &pca9557_cfg, &pca9557_handle));
+    assert(pca9557_handle != NULL);
+    auto pca9557_set_register = [](i2c_master_dev_handle_t pca9557_handle, uint8_t data_addr, uint8_t data) {
+        uint8_t data_[2] = {data_addr, data};
+        ESP_ERROR_CHECK(i2c_master_transmit(pca9557_handle, data_, 2, 50));
+    };
+    pca9557_set_register(pca9557_handle, 0x03, 0xfd);
+    pca9557_set_register(pca9557_handle, 0x01, 0x02);
+#endif
+
+    // Do initialize of related interface: data_if, ctrl_if and gpio_if
+    audio_codec_i2s_cfg_t i2s_cfg = {
+        .port = I2S_NUM_0,
+        .rx_handle = rx_handle_,
+        .tx_handle = tx_handle_,
+    };
+    data_if_ = audio_codec_new_i2s_data(&i2s_cfg);
+    assert(data_if_ != NULL);
+
+    // Output
+    audio_codec_i2c_cfg_t i2c_cfg = {
+        .port = I2C_NUM_1,
+        .addr = AUDIO_CODEC_ES8311_ADDR,
+        .bus_handle = i2c_master_handle_,
+    };
+    out_ctrl_if_ = audio_codec_new_i2c_ctrl(&i2c_cfg);
+    assert(out_ctrl_if_ != NULL);
+
+    gpio_if_ = audio_codec_new_gpio();
+    assert(gpio_if_ != NULL);
+
+    es8311_codec_cfg_t es8311_cfg = {};
+    es8311_cfg.ctrl_if = out_ctrl_if_;
+    es8311_cfg.gpio_if = gpio_if_;
+    es8311_cfg.codec_mode = ESP_CODEC_DEV_WORK_MODE_DAC;
+    es8311_cfg.pa_pin = AUDIO_CODEC_PA_PIN;
+    es8311_cfg.use_mclk = true;
+    es8311_cfg.hw_gain.pa_voltage = 5.0;
+    es8311_cfg.hw_gain.codec_dac_voltage = 3.3;
+    out_codec_if_ = es8311_codec_new(&es8311_cfg);
+    assert(out_codec_if_ != NULL);
+
+    esp_codec_dev_cfg_t dev_cfg = {
+        .dev_type = ESP_CODEC_DEV_TYPE_OUT,
+        .codec_if = out_codec_if_,
+        .data_if = data_if_,
+    };
+    output_dev_ = esp_codec_dev_new(&dev_cfg);
+    assert(output_dev_ != NULL);
+
+    // Input
+    i2c_cfg.addr = AUDIO_CODEC_ES7210_ADDR;
+    in_ctrl_if_ = audio_codec_new_i2c_ctrl(&i2c_cfg);
+    assert(in_ctrl_if_ != NULL);
+
+    es7210_codec_cfg_t es7210_cfg = {};
+    es7210_cfg.ctrl_if = in_ctrl_if_;
+    es7210_cfg.mic_selected = ES7120_SEL_MIC1 | ES7120_SEL_MIC2 | ES7120_SEL_MIC3 | ES7120_SEL_MIC4;
+    in_codec_if_ = es7210_codec_new(&es7210_cfg);
+    assert(in_codec_if_ != NULL);
+
+    dev_cfg.dev_type = ESP_CODEC_DEV_TYPE_IN;
+    dev_cfg.codec_if = in_codec_if_;
+    input_dev_ = esp_codec_dev_new(&dev_cfg);
+    assert(input_dev_ != NULL);
+
+    ESP_LOGI(TAG, "BoxAudioDevice initialized");
+}
+
+void BoxAudioDevice::CreateDuplexChannels() {
+    assert(input_sample_rate_ == output_sample_rate_);
+
+    i2s_chan_config_t chan_cfg = {
+        .id = I2S_NUM_0,
+        .role = I2S_ROLE_MASTER,
+        .dma_desc_num = 6,
+        .dma_frame_num = 240,
+        .auto_clear_after_cb = true,
+        .auto_clear_before_cb = false,
+        .intr_priority = 0,
+    };
+    ESP_ERROR_CHECK(i2s_new_channel(&chan_cfg, &tx_handle_, &rx_handle_));
+
+    i2s_std_config_t std_cfg = {
+        .clk_cfg = {
+            .sample_rate_hz = (uint32_t)output_sample_rate_,
+            .clk_src = I2S_CLK_SRC_DEFAULT,
+            .ext_clk_freq_hz = 0,
+            .mclk_multiple = I2S_MCLK_MULTIPLE_256
+        },
+        .slot_cfg = {
+            .data_bit_width = I2S_DATA_BIT_WIDTH_16BIT,
+            .slot_bit_width = I2S_SLOT_BIT_WIDTH_AUTO,
+            .slot_mode = I2S_SLOT_MODE_STEREO,
+            .slot_mask = I2S_STD_SLOT_BOTH,
+            .ws_width = I2S_DATA_BIT_WIDTH_16BIT,
+            .ws_pol = false,
+            .bit_shift = true,
+            .left_align = true,
+            .big_endian = false,
+            .bit_order_lsb = false
+        },
+        .gpio_cfg = {
+            .mclk = (gpio_num_t)AUDIO_I2S_GPIO_MCLK,
+            .bclk = (gpio_num_t)AUDIO_I2S_GPIO_BCLK,
+            .ws = (gpio_num_t)AUDIO_I2S_GPIO_LRCK,
+            .dout = (gpio_num_t)AUDIO_I2S_GPIO_DOUT,
+            .din = I2S_GPIO_UNUSED,
+            .invert_flags = {
+                .mclk_inv = false,
+                .bclk_inv = false,
+                .ws_inv = false
+            }
+        }
+    };
+
+    i2s_tdm_config_t tdm_cfg = {
+        .clk_cfg = {
+            .sample_rate_hz = (uint32_t)input_sample_rate_,
+            .clk_src = I2S_CLK_SRC_DEFAULT,
+            .ext_clk_freq_hz = 0,
+            .mclk_multiple = I2S_MCLK_MULTIPLE_256,
+            .bclk_div = 8,
+        },
+        .slot_cfg = {
+            .data_bit_width = I2S_DATA_BIT_WIDTH_16BIT,
+            .slot_bit_width = I2S_SLOT_BIT_WIDTH_AUTO,
+            .slot_mode = I2S_SLOT_MODE_STEREO,
+            .slot_mask = i2s_tdm_slot_mask_t(I2S_TDM_SLOT0 | I2S_TDM_SLOT1 | I2S_TDM_SLOT2 | I2S_TDM_SLOT3),
+            .ws_width = I2S_TDM_AUTO_WS_WIDTH,
+            .ws_pol = false,
+            .bit_shift = true,
+            .left_align = false,
+            .big_endian = false,
+            .bit_order_lsb = false,
+            .skip_mask = false,
+            .total_slot = I2S_TDM_AUTO_SLOT_NUM
+        },
+        .gpio_cfg = {
+            .mclk = (gpio_num_t)AUDIO_I2S_GPIO_MCLK,
+            .bclk = (gpio_num_t)AUDIO_I2S_GPIO_BCLK,
+            .ws = (gpio_num_t)AUDIO_I2S_GPIO_LRCK,
+            .dout = I2S_GPIO_UNUSED,
+            .din = (gpio_num_t)AUDIO_I2S_GPIO_DIN,
+            .invert_flags = {
+                .mclk_inv = false,
+                .bclk_inv = false,
+                .ws_inv = false
+            }
+        }
+    };
+
+    ESP_ERROR_CHECK(i2s_channel_init_std_mode(tx_handle_, &std_cfg));
+    ESP_ERROR_CHECK(i2s_channel_init_tdm_mode(rx_handle_, &tdm_cfg));
+    ESP_ERROR_CHECK(i2s_channel_enable(tx_handle_));
+    ESP_ERROR_CHECK(i2s_channel_enable(rx_handle_));
+    ESP_LOGI(TAG, "Duplex channels created");
+}
+
+int BoxAudioDevice::Read(int16_t *buffer, int samples) {
+    if (input_enabled_) {
+        ESP_ERROR_CHECK_WITHOUT_ABORT(esp_codec_dev_read(input_dev_, (void*)buffer, samples * sizeof(int16_t)));
+    }
+    return samples;
+}
+
+int BoxAudioDevice::Write(const int16_t *buffer, int samples) {
+    if (output_enabled_) {
+        ESP_ERROR_CHECK_WITHOUT_ABORT(esp_codec_dev_write(output_dev_, (void*)buffer, samples * sizeof(int16_t)));
+    }
+    return samples;
+}
+
+void BoxAudioDevice::SetOutputVolume(int volume) {
+    ESP_ERROR_CHECK(esp_codec_dev_set_out_vol(output_dev_, volume));
+    AudioDevice::SetOutputVolume(volume);
+}
+
+void BoxAudioDevice::EnableInput(bool enable) {
+    if (enable == input_enabled_) {
+        return;
+    }
+    if (enable) {
+        esp_codec_dev_sample_info_t fs = {
+            .bits_per_sample = 16,
+            .channel = 4,
+            .channel_mask = ESP_CODEC_DEV_MAKE_CHANNEL_MASK(0),
+            .sample_rate = (uint32_t)output_sample_rate_,
+            .mclk_multiple = 0,
+        };
+        if (input_reference_) {
+            fs.channel_mask |= ESP_CODEC_DEV_MAKE_CHANNEL_MASK(1);
+        }
+        ESP_ERROR_CHECK(esp_codec_dev_open(input_dev_, &fs));
+        ESP_ERROR_CHECK(esp_codec_dev_set_in_channel_gain(input_dev_, ESP_CODEC_DEV_MAKE_CHANNEL_MASK(0), 30.0));
+    } else {
+        ESP_ERROR_CHECK(esp_codec_dev_close(input_dev_));
+    }
+    AudioDevice::EnableInput(enable);
+}
+
+void BoxAudioDevice::EnableOutput(bool enable) {
+    if (enable == output_enabled_) {
+        return;
+    }
+    if (enable) {
+        // Play 16bit 1 channel
+        esp_codec_dev_sample_info_t fs = {
+            .bits_per_sample = 16,
+            .channel = 1,
+            .channel_mask = 0,
+            .sample_rate = (uint32_t)output_sample_rate_,
+            .mclk_multiple = 0,
+        };
+        ESP_ERROR_CHECK(esp_codec_dev_open(output_dev_, &fs));
+        ESP_ERROR_CHECK(esp_codec_dev_set_out_vol(output_dev_, output_volume_));
+    } else {
+        ESP_ERROR_CHECK(esp_codec_dev_close(output_dev_));
+    }
+    AudioDevice::EnableOutput(enable);
+}
--- a/main/BoxAudioDevice.h
+++ b/main/BoxAudioDevice.h
@@ -0,0 +1,38 @@
+#ifndef _BOX_AUDIO_DEVICE_H
+#define _BOX_AUDIO_DEVICE_H
+
+#include "AudioDevice.h"
+#include <driver/i2c_master.h>
+#include <driver/i2s_tdm.h>
+#include <esp_codec_dev.h>
+#include <esp_codec_dev_defaults.h>
+
+
+class BoxAudioDevice : public AudioDevice {
+public:
+    BoxAudioDevice();
+    virtual ~BoxAudioDevice();
+    virtual void Initialize() override;
+    virtual void SetOutputVolume(int volume) override;
+    virtual void EnableInput(bool enable) override;
+    virtual void EnableOutput(bool enable) override;
+
+private:
+    i2c_master_bus_handle_t i2c_master_handle_ = nullptr;
+
+    const audio_codec_data_if_t* data_if_ = nullptr;
+    const audio_codec_ctrl_if_t* out_ctrl_if_ = nullptr;
+    const audio_codec_if_t* out_codec_if_ = nullptr;
+    const audio_codec_ctrl_if_t* in_ctrl_if_ = nullptr;
+    const audio_codec_if_t* in_codec_if_ = nullptr;
+    const audio_codec_gpio_if_t* gpio_if_ = nullptr;
+
+    esp_codec_dev_handle_t output_dev_ = nullptr;
+    esp_codec_dev_handle_t input_dev_ = nullptr;
+
+    void CreateDuplexChannels() override;
+    int Read(int16_t* dest, int samples) override;
+    int Write(const int16_t* data, int samples) override;
+};
+
+#endif // _BOX_AUDIO_DEVICE_H
--- a/main/BuiltinLed.cc
+++ b/main/BuiltinLed.cc
@@ -0,0 +1,137 @@
+#include "BuiltinLed.h"
+#include "Board.h"
+#include <cstring>
+#include <esp_log.h>
+
+#define TAG "builtin_led"
+
+BuiltinLed::BuiltinLed() {
+    mutex_ = xSemaphoreCreateMutex();
+    blink_event_group_ = xEventGroupCreate();
+    xEventGroupSetBits(blink_event_group_, BLINK_TASK_STOPPED_BIT);
+
+    if (BUILTIN_LED_GPIO == GPIO_NUM_NC) {
+        ESP_LOGI(TAG, "Builtin LED not connected");
+        return;
+    }
+    Initialize();
+    SetGrey();
+}
+
+BuiltinLed::~BuiltinLed() {
+    StopBlinkInternal();
+    if (led_strip_ != nullptr) {
+        led_strip_del(led_strip_);
+    }
+    if (mutex_ != nullptr) {
+        vSemaphoreDelete(mutex_);
+    }
+    if (blink_event_group_ != nullptr) {
+        vEventGroupDelete(blink_event_group_);
+    }
+}
+
+BuiltinLed& BuiltinLed::GetInstance() {
+    static BuiltinLed instance;
+    return instance;
+}
+
+void BuiltinLed::Initialize() {
+    led_strip_config_t strip_config = {};
+    strip_config.strip_gpio_num = BUILTIN_LED_GPIO;
+    strip_config.max_leds = 1;
+    strip_config.led_pixel_format = LED_PIXEL_FORMAT_GRB;
+    strip_config.led_model = LED_MODEL_WS2812;
+
+    led_strip_rmt_config_t rmt_config = {};
+    rmt_config.resolution_hz = 10 * 1000 * 1000; // 10MHz
+
+    ESP_ERROR_CHECK(led_strip_new_rmt_device(&strip_config, &rmt_config, &led_strip_));
+    led_strip_clear(led_strip_);
+}
+
+void BuiltinLed::SetColor(uint8_t r, uint8_t g, uint8_t b) {
+    r_ = r;
+    g_ = g;
+    b_ = b;
+}
+
+void BuiltinLed::TurnOn() {
+    if (led_strip_ == nullptr) {
+        return;
+    }
+    StopBlinkInternal();
+    xSemaphoreTake(mutex_, portMAX_DELAY);
+    led_strip_set_pixel(led_strip_, 0, r_, g_, b_);
+    led_strip_refresh(led_strip_);
+    xSemaphoreGive(mutex_);
+}
+
+void BuiltinLed::TurnOff() {
+    if (led_strip_ == nullptr) {
+        return;
+    }
+    StopBlinkInternal();
+    xSemaphoreTake(mutex_, portMAX_DELAY);
+    led_strip_clear(led_strip_);
+    xSemaphoreGive(mutex_);
+}
+
+void BuiltinLed::BlinkOnce() {
+    Blink(1, 100);
+}
+
+void BuiltinLed::Blink(int times, int interval_ms) {
+    StartBlinkTask(times, interval_ms);
+}
+
+void BuiltinLed::StartContinuousBlink(int interval_ms) {
+    StartBlinkTask(BLINK_INFINITE, interval_ms);
+}
+
+void BuiltinLed::StartBlinkTask(int times, int interval_ms) {
+    if (led_strip_ == nullptr) {
+        return;
+    }
+    StopBlinkInternal();
+    xSemaphoreTake(mutex_, portMAX_DELAY);
+    
+    blink_times_ = times;
+    blink_interval_ms_ = interval_ms;
+    should_blink_ = true;
+
+    xEventGroupClearBits(blink_event_group_, BLINK_TASK_STOPPED_BIT);
+    xEventGroupSetBits(blink_event_group_, BLINK_TASK_RUNNING_BIT);
+
+    xTaskCreate([](void* obj) {
+        auto this_ = static_cast<BuiltinLed*>(obj);
+        int count = 0;
+        while (this_->should_blink_ && (this_->blink_times_ == BLINK_INFINITE || count < this_->blink_times_)) {
+            xSemaphoreTake(this_->mutex_, portMAX_DELAY);
+            led_strip_set_pixel(this_->led_strip_, 0, this_->r_, this_->g_, this_->b_);
+            led_strip_refresh(this_->led_strip_);
+            xSemaphoreGive(this_->mutex_);
+
+            vTaskDelay(this_->blink_interval_ms_ / portTICK_PERIOD_MS);
+            if (!this_->should_blink_) break;
+
+            xSemaphoreTake(this_->mutex_, portMAX_DELAY);
+            led_strip_clear(this_->led_strip_);
+            xSemaphoreGive(this_->mutex_);
+
+            vTaskDelay(this_->blink_interval_ms_ / portTICK_PERIOD_MS);
+            if (this_->blink_times_ != BLINK_INFINITE) count++;
+        }
+        this_->blink_task_ = nullptr;
+        xEventGroupClearBits(this_->blink_event_group_, BLINK_TASK_RUNNING_BIT);
+        xEventGroupSetBits(this_->blink_event_group_, BLINK_TASK_STOPPED_BIT);
+        vTaskDelete(NULL);
+    }, "blink", 2048, this, tskIDLE_PRIORITY, &blink_task_);
+
+    xSemaphoreGive(mutex_);
+}
+
+void BuiltinLed::StopBlinkInternal() {
+    should_blink_ = false;
+    xEventGroupWaitBits(blink_event_group_, BLINK_TASK_STOPPED_BIT, pdFALSE, pdTRUE, portMAX_DELAY);
+}
--- a/main/BuiltinLed.h
+++ b/main/BuiltinLed.h
@@ -0,0 +1,52 @@
+#ifndef _BUILTIN_LED_H_
+#define _BUILTIN_LED_H_
+
+#include <led_strip.h>
+#include <freertos/semphr.h>
+#include <freertos/task.h>
+#include <freertos/event_groups.h>
+#include <atomic>
+
+#define BLINK_INFINITE -1
+#define BLINK_TASK_STOPPED_BIT BIT0
+#define BLINK_TASK_RUNNING_BIT BIT1
+
+#define DEFAULT_BRIGHTNESS 16
+
+class BuiltinLed {
+public:
+    static BuiltinLed& GetInstance();
+
+    void BlinkOnce();
+    void Blink(int times, int interval_ms);
+    void StartContinuousBlink(int interval_ms);
+    void TurnOn();
+    void TurnOff();
+    void SetColor(uint8_t r, uint8_t g, uint8_t b);
+    void SetWhite(uint8_t brightness = DEFAULT_BRIGHTNESS) { SetColor(brightness, brightness, brightness); }
+    void SetGrey(uint8_t brightness = DEFAULT_BRIGHTNESS) { SetColor(brightness, brightness, brightness); }
+    void SetRed(uint8_t brightness = DEFAULT_BRIGHTNESS) { SetColor(brightness, 0, 0); }
+    void SetGreen(uint8_t brightness = DEFAULT_BRIGHTNESS) { SetColor(0, brightness, 0); }
+    void SetBlue(uint8_t brightness = DEFAULT_BRIGHTNESS) { SetColor(0, 0, brightness); }
+
+private:
+    BuiltinLed();
+    ~BuiltinLed();
+    BuiltinLed(const BuiltinLed&) = delete;
+    BuiltinLed& operator=(const BuiltinLed&) = delete;
+
+    SemaphoreHandle_t mutex_;
+    EventGroupHandle_t blink_event_group_;
+    TaskHandle_t blink_task_ = nullptr;
+    led_strip_handle_t led_strip_ = nullptr;
+    uint8_t r_ = 0, g_ = 0, b_ = 0;
+    int blink_times_ = 0;
+    int blink_interval_ms_ = 0;
+    std::atomic<bool> should_blink_{false};
+
+    void Initialize();
+    void StartBlinkTask(int times, int interval_ms);
+    void StopBlinkInternal();
+};
+
+#endif // _BUILTIN_LED_H_
--- a/main/Button.cc
+++ b/main/Button.cc
@@ -4,6 +4,9 @@
 static const char* TAG = "Button";

 Button::Button(gpio_num_t gpio_num) : gpio_num_(gpio_num) {
+    if (gpio_num == GPIO_NUM_NC) {
+        return;
+    }
    button_config_t button_config = {
        .type = BUTTON_TYPE_GPIO,
        .long_press_time = 1000,
@@ -27,6 +30,9 @@ Button::~Button() {
 }

 void Button::OnPress(std::function<void()> callback) {
+    if (button_handle_ == nullptr) {
+        return;
+    }
    on_press_ = callback;
    iot_button_register_cb(button_handle_, BUTTON_PRESS_DOWN, [](void* handle, void* usr_data) {
        Button* button = static_cast<Button*>(usr_data);
@@ -37,6 +43,9 @@ void Button::OnPress(std::function<void()> callback) {
 }

 void Button::OnLongPress(std::function<void()> callback) {
+    if (button_handle_ == nullptr) {
+        return;
+    }
    on_long_press_ = callback;
    iot_button_register_cb(button_handle_, BUTTON_LONG_PRESS_START, [](void* handle, void* usr_data) {
        Button* button = static_cast<Button*>(usr_data);
@@ -47,6 +56,9 @@ void Button::OnLongPress(std::function<void()> callback) {
 }

 void Button::OnClick(std::function<void()> callback) {
+    if (button_handle_ == nullptr) {
+        return;
+    }
    on_click_ = callback;
    iot_button_register_cb(button_handle_, BUTTON_SINGLE_CLICK, [](void* handle, void* usr_data) {
        Button* button = static_cast<Button*>(usr_data);
@@ -57,6 +69,9 @@ void Button::OnClick(std::function<void()> callback) {
 }

 void Button::OnDoubleClick(std::function<void()> callback) {
+    if (button_handle_ == nullptr) {
+        return;
+    }
    on_double_click_ = callback;
    iot_button_register_cb(button_handle_, BUTTON_DOUBLE_CLICK, [](void* handle, void* usr_data) {
        Button* button = static_cast<Button*>(usr_data);
--- a/main/CMakeLists.txt
+++ b/main/CMakeLists.txt
@@ -3,15 +3,65 @@ set(SOURCES "AudioDevice.cc"
            "SystemInfo.cc"
            "SystemReset.cc"
            "Application.cc"
-            "Display.cc"
            "Button.cc"
+            "BuiltinLed.cc"
+            "Display.cc"
+            "Board.cc"
            "main.cc"
            )
+set(INCLUDE_DIRS ".")
+
+# 根据 BOARD_TYPE 配置添加对应的板级文件
+if(CONFIG_BOARD_TYPE_BREAD_COMPACT_WIFI)
+    # add all files from boards/bread-compact-wifi
+    set(BOARD_TYPE "bread-compact-wifi")
+    file(GLOB BOARD_SOURCES ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE}/*.cc)
+    list(APPEND SOURCES ${BOARD_SOURCES} "WifiBoard.cc")
+    list(APPEND INCLUDE_DIRS ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE})
+elseif(CONFIG_BOARD_TYPE_BREAD_COMPACT_ML307)
+    # add all files from boards/bread-compact-ml307
+    set(BOARD_TYPE "bread-compact-ml307")
+    file(GLOB BOARD_SOURCES ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE}/*.cc)
+    list(APPEND SOURCES ${BOARD_SOURCES} "Ml307Board.cc")
+    list(APPEND INCLUDE_DIRS ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE})
+elseif(CONFIG_BOARD_TYPE_ESP_BOX_3)
+    # add all files from boards/esp-box-3
+    set(BOARD_TYPE "esp-box-3")
+    file(GLOB BOARD_SOURCES ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE}/*.cc)
+    list(APPEND SOURCES ${BOARD_SOURCES} "WifiBoard.cc")
+    list(APPEND INCLUDE_DIRS ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE})
+    list(APPEND SOURCES "BoxAudioDevice.cc")
+elseif(CONFIG_BOARD_TYPE_KEVIN_BOX_0)
+    # add all files from boards/kevin-box-0
+    set(BOARD_TYPE "kevin-box-0")
+    file(GLOB BOARD_SOURCES ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE}/*.cc)
+    list(APPEND SOURCES ${BOARD_SOURCES} "Ml307Board.cc")
+    list(APPEND INCLUDE_DIRS ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE})
+    list(APPEND SOURCES "BoxAudioDevice.cc")
+elseif(CONFIG_BOARD_TYPE_KEVIN_BOX_1)
+    # add all files from boards/kevin-box-1
+    set(BOARD_TYPE "kevin-box-1")
+    file(GLOB BOARD_SOURCES ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE}/*.cc)
+    list(APPEND SOURCES ${BOARD_SOURCES} "Ml307Board.cc")
+    list(APPEND INCLUDE_DIRS ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE})
+    list(APPEND SOURCES "BoxAudioDevice.cc")
+elseif(CONFIG_BOARD_TYPE_LICHUANG_DEV)
+    # add all files from boards/lichuang-dev
+    set(BOARD_TYPE "lichuang-dev")
+    file(GLOB BOARD_SOURCES ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE}/*.cc)
+    list(APPEND SOURCES ${BOARD_SOURCES} "WifiBoard.cc")
+    list(APPEND INCLUDE_DIRS ${CMAKE_CURRENT_SOURCE_DIR}/boards/${BOARD_TYPE})
+    list(APPEND SOURCES "BoxAudioDevice.cc")
+endif()

 if(CONFIG_USE_AFE_SR)
    list(APPEND SOURCES "AudioProcessor.cc" "WakeWordDetect.cc")
 endif()

 idf_component_register(SRCS ${SOURCES}
-                    INCLUDE_DIRS "."
+                    EMBED_FILES "assets/err_reg.p3" "assets/err_pin.p3" "assets/err_wificonfig.p3"
+                    INCLUDE_DIRS ${INCLUDE_DIRS}
                    )
+
+# 使用 target_compile_definitions 来定义 BOARD_TYPE
+target_compile_definitions(${COMPONENT_LIB} PRIVATE BOARD_TYPE=\"${BOARD_TYPE}\")
--- a/main/Display.cc
+++ b/main/Display.cc
@@ -1,6 +1,3 @@
-
-#include "Display.h"
-
 #include <esp_log.h>
 #include <esp_err.h>
 #include <esp_lcd_panel_ops.h>
@@ -9,12 +6,17 @@
 #include <string>
 #include <cstdlib>

+#include "Display.h"
+#include "Board.h"
+#include "Application.h"
+
 #define TAG "Display"

-#ifdef CONFIG_USE_DISPLAY
-
 Display::Display(int sda_pin, int scl_pin) : sda_pin_(sda_pin), scl_pin_(scl_pin) {
-    ESP_LOGI(TAG, "Display Pins: %d, %d", sda_pin_, scl_pin_);
+    if (sda_pin_ == GPIO_NUM_NC || scl_pin_ == GPIO_NUM_NC) {
+        ESP_LOGI(TAG, "Display not connected");
+        return;
+    }

    i2c_master_bus_config_t bus_config = {
        .i2c_port = I2C_NUM_0,
@@ -22,7 +24,7 @@ Display::Display(int sda_pin, int scl_pin) : sda_pin_(sda_pin), scl_pin_(scl_pin
        .scl_io_num = (gpio_num_t)scl_pin_,
        .clk_source = I2C_CLK_SRC_DEFAULT,
        .glitch_ignore_cnt = 7,
-        .intr_priority = 1,
+        .intr_priority = 0,
        .trans_queue_depth = 0,
        .flags = {
            .enable_internal_pullup = 1,
@@ -44,7 +46,7 @@ Display::Display(int sda_pin, int scl_pin) : sda_pin_(sda_pin), scl_pin_(scl_pin
            .dc_low_on_data = 0,
            .disable_control_phase = 0,
        },
-        .scl_speed_hz = 400 * 1000,
+        .scl_speed_hz = 100 * 1000,
    };

    ESP_ERROR_CHECK(esp_lcd_new_panel_io_i2c_v2(i2c_bus_, &io_config, &panel_io_));
@@ -55,7 +57,7 @@ Display::Display(int sda_pin, int scl_pin) : sda_pin_(sda_pin), scl_pin_(scl_pin
    panel_config.bits_per_pixel = 1;

    esp_lcd_panel_ssd1306_config_t ssd1306_config = {
-        .height = CONFIG_DISPLAY_HEIGHT
+        .height = DISPLAY_HEIGHT
    };
    panel_config.vendor_config = &ssd1306_config;

@@ -73,45 +75,66 @@ Display::Display(int sda_pin, int scl_pin) : sda_pin_(sda_pin), scl_pin_(scl_pin
    lvgl_port_cfg_t port_cfg = ESP_LVGL_PORT_INIT_CONFIG();
    lvgl_port_init(&port_cfg);

-    const lvgl_port_display_cfg_t display_cfg = {
-        .io_handle = panel_io_,
-        .panel_handle = panel_,
-        .buffer_size = 128 * CONFIG_DISPLAY_HEIGHT,
-        .double_buffer = true,
-        .hres = 128,
-        .vres = CONFIG_DISPLAY_HEIGHT,
-        .monochrome = true,
-        .rotation = {
-            .swap_xy = 0,
-            .mirror_x = 0,
-            .mirror_y = 0,
-        },
-        .flags = {
-            .buff_dma = 0,
-            .buff_spiram = 0,
-        },
-    };
-    disp_ = lvgl_port_add_disp(&display_cfg);
-    lv_disp_set_rotation(disp_, LV_DISP_ROT_180);
-
    // Set the display to on
    ESP_LOGI(TAG, "Turning display on");
    ESP_ERROR_CHECK(esp_lcd_panel_disp_on_off(panel_, true));
+
+    ESP_LOGI(TAG, "Adding LCD screen");
+    const lvgl_port_display_cfg_t display_cfg = {
+        .io_handle = panel_io_,
+        .panel_handle = panel_,
+        .control_handle = nullptr,
+        .buffer_size = DISPLAY_WIDTH * DISPLAY_HEIGHT,
+        .double_buffer = false,
+        .trans_size = 0,
+        .hres = DISPLAY_WIDTH,
+        .vres = DISPLAY_HEIGHT,
+        .monochrome = true,
+        .rotation = {
+            .swap_xy = false,
+            .mirror_x = DISPLAY_MIRROR_X,
+            .mirror_y = DISPLAY_MIRROR_Y,
+        },
+        .flags = {
+            .buff_dma = 1,
+            .buff_spiram = 0,
+            .sw_rotate = 0,
+            .full_refresh = 0,
+            .direct_mode = 0,
+        },
+    };
+    disp_ = lvgl_port_add_disp(&display_cfg);;
    
    ESP_LOGI(TAG, "Display Loading...");
    if (lvgl_port_lock(0)) {
        label_ = lv_label_create(lv_disp_get_scr_act(disp_));
+        // lv_obj_set_style_text_font(label_, font_, 0);
        lv_label_set_text(label_, "Initializing...");
        lv_obj_set_width(label_, disp_->driver->hor_res);
        lv_obj_set_height(label_, disp_->driver->ver_res);

        notification_ = lv_label_create(lv_disp_get_scr_act(disp_));
+        // lv_obj_set_style_text_font(notification_, font_, 0);
        lv_label_set_text(notification_, "Notification\nTest");
        lv_obj_set_width(notification_, disp_->driver->hor_res);
        lv_obj_set_height(notification_, disp_->driver->ver_res);
        lv_obj_set_style_opa(notification_, LV_OPA_MIN, 0);
        lvgl_port_unlock();
    }
+
+    // Create a timer to update the display every 10 seconds
+    esp_timer_create_args_t update_display_timer_args = {
+        .callback = [](void *arg) {
+            Display* display = static_cast<Display*>(arg);
+            display->UpdateDisplay();
+        },
+        .arg = this,
+        .dispatch_method = ESP_TIMER_TASK,
+        .name = "UpdateDisplay",
+        .skip_unhandled_events = false,
+    };
+    ESP_ERROR_CHECK(esp_timer_create(&update_display_timer_args, &update_display_timer_));
+    ESP_ERROR_CHECK(esp_timer_start_periodic(update_display_timer_, 10 * 1000000));
 }

 Display::~Display() {
@@ -119,6 +142,10 @@ Display::~Display() {
        esp_timer_stop(notification_timer_);
        esp_timer_delete(notification_timer_);
    }
+    if (update_display_timer_ != nullptr) {
+        esp_timer_stop(update_display_timer_);
+        esp_timer_delete(update_display_timer_);
+    }

    lvgl_port_lock(0);
    if (label_ != nullptr) {
@@ -127,6 +154,10 @@ Display::~Display() {
    }
    lvgl_port_unlock();

+    if (font_ != nullptr) {
+        lv_font_free(font_);
+    }
+
    if (disp_ != nullptr) {
        lvgl_port_deinit();
        esp_lcd_panel_del(panel_);
@@ -176,4 +207,31 @@ void Display::ShowNotification(const std::string &text) {
    }
 }

-#endif
+void Display::UpdateDisplay() {
+    auto chat_state = Application::GetInstance().GetChatState();
+    if (chat_state == kChatStateIdle || chat_state == kChatStateConnecting || chat_state == kChatStateListening) {
+        std::string text;
+        auto& board = Board::GetInstance();
+        std::string network_name;
+        int signal_quality;
+        std::string signal_quality_text;
+        if (!board.GetNetworkState(network_name, signal_quality, signal_quality_text)) {
+            text = "No network";
+        } else {
+            text = network_name + "\n" + signal_quality_text;
+            if (std::abs(signal_quality) != 99) {
+                text += " (" + std::to_string(signal_quality) + ")";
+            }
+        }
+
+        int battery_voltage;
+        bool charging;
+        if (board.GetBatteryVoltage(battery_voltage, charging)) {
+            text += "\n" + std::to_string(battery_voltage) + "mV";
+            if (charging) {
+                text += " (Charging)";
+            }
+        }
+        SetText(text);
+    }
+}
--- a/main/Display.h
+++ b/main/Display.h
@@ -17,6 +17,8 @@ public:
    void SetText(const std::string &text);
    void ShowNotification(const std::string &text);

+    void UpdateDisplay();
+
 private:
    int sda_pin_;
    int scl_pin_;
@@ -26,9 +28,11 @@ private:
    esp_lcd_panel_io_handle_t panel_io_ = nullptr;
    esp_lcd_panel_handle_t panel_ = nullptr;
    lv_disp_t *disp_ = nullptr;
+    lv_font_t *font_ = nullptr;
    lv_obj_t *label_ = nullptr;
    lv_obj_t *notification_ = nullptr;
    esp_timer_handle_t notification_timer_ = nullptr;
+    esp_timer_handle_t update_display_timer_ = nullptr;

    std::string text_;
 };
--- a/main/FirmwareUpgrade.cc
+++ b/main/FirmwareUpgrade.cc
@@ -1,11 +1,14 @@
 #include "FirmwareUpgrade.h"
 #include "SystemInfo.h"
+#include "Board.h"
+
 #include <cJSON.h>
 #include <esp_log.h>
 #include <esp_partition.h>
 #include <esp_http_client.h>
 #include <esp_ota_ops.h>
 #include <esp_app_format.h>
+#include <esp_chip_info.h>

 #include <vector>
 #include <sstream>
@@ -14,7 +17,7 @@
 #define TAG "FirmwareUpgrade"


-FirmwareUpgrade::FirmwareUpgrade(Http& http) : http_(http) {
+FirmwareUpgrade::FirmwareUpgrade() {
 }

 FirmwareUpgrade::~FirmwareUpgrade() {
@@ -24,10 +27,6 @@ void FirmwareUpgrade::SetCheckVersionUrl(std::string check_version_url) {
    check_version_url_ = check_version_url;
 }

-void FirmwareUpgrade::SetPostData(const std::string& post_data) {
-    post_data_ = post_data;
-}
-
 void FirmwareUpgrade::SetHeader(const std::string& key, const std::string& value) {
    headers_[key] = value;
 }
@@ -41,20 +40,18 @@ void FirmwareUpgrade::CheckVersion() {
        return;
    }

+    auto http = Board::GetInstance().CreateHttp();
    for (const auto& header : headers_) {
-        http_.SetHeader(header.first, header.second);
+        http->SetHeader(header.first, header.second);
    }

-    if (post_data_.empty()) {
-        http_.Open("GET", check_version_url_);
-    } else {
-        http_.SetHeader("Content-Type", "application/json");
-        http_.SetContent(post_data_);
-        http_.Open("POST", check_version_url_);
-    }
+    http->SetHeader("Content-Type", "application/json");
+    http->SetContent(GetPostData());
+    http->Open("POST", check_version_url_);

-    auto response = http_.GetBody();
-    http_.Close();
+    auto response = http->GetBody();
+    http->Close();
+    delete http;

    // Response: { "firmware": { "version": "1.0.0", "url": "http://" } }
    // Parse the JSON response and check if the version is newer
@@ -130,15 +127,17 @@ void FirmwareUpgrade::Upgrade(const std::string& firmware_url) {
    bool image_header_checked = false;
    std::string image_header;

-    if (!http_.Open("GET", firmware_url)) {
+    auto http = Board::GetInstance().CreateHttp();
+    if (!http->Open("GET", firmware_url)) {
        ESP_LOGE(TAG, "Failed to open HTTP connection");
+        delete http;
        return;
    }

-    size_t content_length = http_.GetBodyLength();
+    size_t content_length = http->GetBodyLength();
    if (content_length == 0) {
        ESP_LOGE(TAG, "Failed to get content length");
-        http_.Close();
+        delete http;
        return;
    }

@@ -146,10 +145,10 @@ void FirmwareUpgrade::Upgrade(const std::string& firmware_url) {
    size_t total_read = 0, recent_read = 0;
    auto last_calc_time = esp_timer_get_time();
    while (true) {
-        int ret = http_.Read(buffer, sizeof(buffer));
+        int ret = http->Read(buffer, sizeof(buffer));
        if (ret < 0) {
            ESP_LOGE(TAG, "Failed to read HTTP data: %s", esp_err_to_name(ret));
-            http_.Close();
+            delete http;
            return;
        }

@@ -181,13 +180,13 @@ void FirmwareUpgrade::Upgrade(const std::string& firmware_url) {
                auto current_version = esp_app_get_description()->version;
                if (memcmp(new_app_info.version, current_version, sizeof(new_app_info.version)) == 0) {
                    ESP_LOGE(TAG, "Firmware version is the same, skipping upgrade");
-                    http_.Close();
+                    delete http;
                    return;
                }

                if (esp_ota_begin(update_partition, OTA_WITH_SEQUENTIAL_WRITES, &update_handle)) {
                    esp_ota_abort(update_handle);
-                    http_.Close();
+                    delete http;
                    ESP_LOGE(TAG, "Failed to begin OTA");
                    return;
                }
@@ -199,11 +198,11 @@ void FirmwareUpgrade::Upgrade(const std::string& firmware_url) {
        if (err != ESP_OK) {
            ESP_LOGE(TAG, "Failed to write OTA data: %s", esp_err_to_name(err));
            esp_ota_abort(update_handle);
-            http_.Close();
+            delete http;
            return;
        }
    }
-    http_.Close();
+    delete http;

    esp_err_t err = esp_ota_end(update_handle);
    if (err != ESP_OK) {
@@ -257,3 +256,99 @@ bool FirmwareUpgrade::IsNewVersionAvailable(const std::string& currentVersion, c
    
    return newer.size() > current.size();
 }
+
+void FirmwareUpgrade::SetBoardJson(const std::string& board_json) {
+    board_json_ = board_json;
+}
+
+std::string FirmwareUpgrade::GetPostData() {
+    /* 
+        {
+            "flash_size": 4194304,
+            "psram_size": 0,
+            "minimum_free_heap_size": 123456,
+            "mac_address": "00:00:00:00:00:00",
+            "chip_model_name": "esp32s3",
+            "chip_info": {
+                "model": 1,
+                "cores": 2,
+                "revision": 0,
+                "features": 0
+            },
+            "application": {
+                "name": "my-app",
+                "version": "1.0.0",
+                "compile_time": "2021-01-01T00:00:00Z"
+                "idf_version": "4.2-dev"
+                "elf_sha256": ""
+            },
+            "partition_table": [
+                "app": {
+                    "label": "app",
+                    "type": 1,
+                    "subtype": 2,
+                    "address": 0x10000,
+                    "size": 0x100000
+                }
+            ],
+            "ota": {
+                "label": "ota_0"
+            }
+        }
+    */
+    std::string json = "{";
+    json += "\"flash_size\":" + std::to_string(SystemInfo::GetFlashSize()) + ",";
+    json += "\"minimum_free_heap_size\":" + std::to_string(SystemInfo::GetMinimumFreeHeapSize()) + ",";
+    json += "\"mac_address\":\"" + SystemInfo::GetMacAddress() + "\",";
+    json += "\"chip_model_name\":\"" + SystemInfo::GetChipModelName() + "\",";
+    json += "\"chip_info\":{";
+
+    esp_chip_info_t chip_info;
+    esp_chip_info(&chip_info);
+    json += "\"model\":" + std::to_string(chip_info.model) + ",";
+    json += "\"cores\":" + std::to_string(chip_info.cores) + ",";
+    json += "\"revision\":" + std::to_string(chip_info.revision) + ",";
+    json += "\"features\":" + std::to_string(chip_info.features);
+    json += "},";
+
+    json += "\"application\":{";
+    auto app_desc = esp_app_get_description();
+    json += "\"name\":\"" + std::string(app_desc->project_name) + "\",";
+    json += "\"version\":\"" + std::string(app_desc->version) + "\",";
+    json += "\"compile_time\":\"" + std::string(app_desc->date) + "T" + std::string(app_desc->time) + "Z\",";
+    json += "\"idf_version\":\"" + std::string(app_desc->idf_ver) + "\",";
+
+    char sha256_str[65];
+    for (int i = 0; i < 32; i++) {
+        snprintf(sha256_str + i * 2, sizeof(sha256_str) - i * 2, "%02x", app_desc->app_elf_sha256[i]);
+    }
+    json += "\"elf_sha256\":\"" + std::string(sha256_str) + "\"";
+    json += "},";
+
+    json += "\"partition_table\": [";
+    esp_partition_iterator_t it = esp_partition_find(ESP_PARTITION_TYPE_ANY, ESP_PARTITION_SUBTYPE_ANY, NULL);
+    while (it) {
+        const esp_partition_t *partition = esp_partition_get(it);
+        json += "{";
+        json += "\"label\":\"" + std::string(partition->label) + "\",";
+        json += "\"type\":" + std::to_string(partition->type) + ",";
+        json += "\"subtype\":" + std::to_string(partition->subtype) + ",";
+        json += "\"address\":" + std::to_string(partition->address) + ",";
+        json += "\"size\":" + std::to_string(partition->size);
+        json += "},";
+        it = esp_partition_next(it);
+    }
+    json.pop_back(); // Remove the last comma
+    json += "],";
+
+    json += "\"ota\":{";
+    auto ota_partition = esp_ota_get_running_partition();
+    json += "\"label\":\"" + std::string(ota_partition->label) + "\"";
+    json += "},";
+
+    json += "\"board\":" + board_json_;
+
+    // Close the JSON object
+    json += "}";
+    return json;
+}
--- a/main/FirmwareUpgrade.h
+++ b/main/FirmwareUpgrade.h
@@ -5,15 +5,13 @@
 #include <string>
 #include <map>

-#include <Http.h>
-
 class FirmwareUpgrade {
 public:
-    FirmwareUpgrade(Http& http);
+    FirmwareUpgrade();
    ~FirmwareUpgrade();

+    void SetBoardJson(const std::string& board_json);
    void SetCheckVersionUrl(std::string check_version_url);
-    void SetPostData(const std::string& post_data);
    void SetHeader(const std::string& key, const std::string& value);
    void CheckVersion();
    bool HasNewVersion() { return has_new_version_; }
@@ -21,18 +19,18 @@ public:
    void MarkCurrentVersionValid();

 private:
-    Http& http_;
    std::string check_version_url_;
    bool has_new_version_ = false;
    std::string firmware_version_;
    std::string firmware_url_;
-    std::string post_data_;
+    std::string board_json_;
    std::map<std::string, std::string> headers_;

    void Upgrade(const std::string& firmware_url);
    std::function<void(int progress, size_t speed)> upgrade_callback_;
    std::vector<int> ParseVersion(const std::string& version);
    bool IsNewVersionAvailable(const std::string& currentVersion, const std::string& newVersion);
+    std::string GetPostData();
 };

 #endif // _FIRMWARE_UPGRADE_H
--- a/main/Kconfig.projbuild
+++ b/main/Kconfig.projbuild
@@ -18,73 +18,24 @@ config WEBSOCKET_ACCESS_TOKEN
    help
        Access token for websocket communication.

-config AUDIO_INPUT_SAMPLE_RATE
-    int "Audio Input Sample Rate"
-    default 16000
+choice BOARD_TYPE
+    prompt "Board Type"
+    default BOARD_TYPE_BREAD_COMPACT_WIFI
    help
-        Audio input sample rate.
-
-config AUDIO_OUTPUT_SAMPLE_RATE
-    int "Audio Output Sample Rate"
-    default 24000
-    help
-        Audio output sample rate.
-
-config AUDIO_DEVICE_I2S_MIC_GPIO_WS
-    int "I2S GPIO WS"
-    default 4
-    help
-        GPIO number of the I2S WS.
-
-config AUDIO_DEVICE_I2S_MIC_GPIO_BCLK
-    int "I2S GPIO BCLK"
-    default 5
-    help
-        GPIO number of the I2S BCLK.
-
-config AUDIO_DEVICE_I2S_MIC_GPIO_DIN
-    int "I2S GPIO DIN"
-    default 6
-    help
-        GPIO number of the I2S DIN.
-
-config AUDIO_DEVICE_I2S_SPK_GPIO_DOUT
-    int "I2S GPIO DOUT"
-    default 7
-    help
-        GPIO number of the I2S DOUT.
-    
-config AUDIO_DEVICE_I2S_SIMPLEX
-    bool "I2S Simplex"
-    default y
-    help
-        Enable I2S Simplex mode.
-    
-config AUDIO_DEVICE_I2S_SPK_GPIO_BCLK
-    int "I2S SPK GPIO BCLK"
-    default 15
-    depends on AUDIO_DEVICE_I2S_SIMPLEX
-    help
-        GPIO number of the I2S MIC BCLK.
-    
-config AUDIO_DEVICE_I2S_SPK_GPIO_WS
-    int "I2S SPK GPIO WS"
-    default 16
-    depends on AUDIO_DEVICE_I2S_SIMPLEX
-    help
-        GPIO number of the I2S MIC WS.
-
-config BOOT_BUTTON_GPIO
-    int "Boot Button GPIO"
-    default 0
-    help
-        GPIO number of the boot button.
-
-config VOLUME_UP_BUTTON_GPIO
-    int "Volume Up Button GPIO"
-    default 40
-    help
-        GPIO number of the volume up button.
+        Board type. 开发板类型
+    config BOARD_TYPE_BREAD_COMPACT_WIFI
+        bool "面包板新版接线（WiFi）"
+    config BOARD_TYPE_BREAD_COMPACT_ML307
+        bool "面包板新版接线（ML307 AT）"
+    config BOARD_TYPE_ESP_BOX_3
+        bool "ESP BOX 3"
+    config BOARD_TYPE_KEVIN_BOX_0
+        bool "Kevin Box 0"
+    config BOARD_TYPE_KEVIN_BOX_1
+        bool "Kevin Box 1"
+    config BOARD_TYPE_LICHUANG_DEV
+        bool "立创开发板"
+endchoice

 config USE_AFE_SR
    bool "Use Espressif AFE SR"
@@ -92,51 +43,4 @@ config USE_AFE_SR
    help
        Use AFE SR for wake word detection.

-config USE_ML307
-    bool "Use ML307"
-    default n
-    help
-        Use ML307 as the modem.
-
-config ML307_RX_PIN
-    int "ML307 RX Pin"
-    default 11
-    depends on USE_ML307
-    help
-        GPIO number of the ML307 RX.
-
-config ML307_TX_PIN
-    int "ML307 TX Pin"
-    default 12
-    depends on USE_ML307
-    help
-        GPIO number of the ML307 TX.
-
-config USE_DISPLAY
-    bool "Use Display"
-    default n
-    help
-        Use Display.
-
-config DISPLAY_HEIGHT
-    int "Display Height"
-    default 32
-    depends on USE_DISPLAY
-    help
-        Display height in pixels.
-
-config DISPLAY_SDA_PIN
-    int "Display SDA Pin"
-    default 41
-    depends on USE_DISPLAY
-    help
-        GPIO number of the Display SDA.
-
-config DISPLAY_SCL_PIN
-    int "Display SCL Pin"
-    default 42
-    depends on USE_DISPLAY
-    help
-        GPIO number of the Display SCL.
-
 endmenu
--- a/main/Ml307Board.cc
+++ b/main/Ml307Board.cc
@@ -0,0 +1,112 @@
+#include "Ml307Board.h"
+#include "Application.h"
+
+#include <esp_log.h>
+#include <Ml307Http.h>
+#include <Ml307SslTransport.h>
+#include <WebSocket.h>
+#include <esp_timer.h>
+
+static const char *TAG = "Ml307Board";
+
+static std::string csq_to_string(int csq) {
+    if (csq == -1) {
+        return "No network";
+    } else if (csq >= 0 && csq <= 9) {
+        return "Very bad";
+    } else if (csq >= 10 && csq <= 14) {
+        return "Bad";
+    } else if (csq >= 15 && csq <= 19) {
+        return "Fair";
+    } else if (csq >= 20 && csq <= 24) {
+        return "Good";
+    } else if (csq >= 25 && csq <= 31) {
+        return "Very good";
+    }
+    return "Invalid";
+}
+
+
+Ml307Board::Ml307Board() : modem_(ML307_TX_PIN, ML307_RX_PIN, 4096) {
+}
+
+void Ml307Board::StartNetwork() {
+    auto& application = Application::GetInstance();
+    auto& display = application.GetDisplay();
+    display.SetText(std::string("Wait for network\n"));
+    int result = modem_.WaitForNetworkReady();
+    if (result == -1) {
+        application.Alert("Error", "PIN is not ready");
+    } else if (result == -2) {
+        application.Alert("Error", "Registration denied");
+    }
+
+    // Print the ML307 modem information
+    std::string module_name = modem_.GetModuleName();
+    std::string imei = modem_.GetImei();
+    std::string iccid = modem_.GetIccid();
+    ESP_LOGI(TAG, "ML307 Module: %s", module_name.c_str());
+    ESP_LOGI(TAG, "ML307 IMEI: %s", imei.c_str());
+    ESP_LOGI(TAG, "ML307 ICCID: %s", iccid.c_str());
+}
+
+void Ml307Board::StartModem() {
+    auto& display = Application::GetInstance().GetDisplay();
+    display.SetText(std::string("Starting modem"));
+    modem_.SetDebug(false);
+    modem_.SetBaudRate(921600);
+
+    auto& application = Application::GetInstance();
+    // If low power, the material ready event will be triggered by the modem because of a reset
+    modem_.OnMaterialReady([this, &application]() {
+        ESP_LOGI(TAG, "ML307 material ready");
+        application.Schedule([this, &application]() {
+            application.SetChatState(kChatStateIdle);
+            StartNetwork();
+        });
+    });
+}
+
+void Ml307Board::Initialize() {
+    ESP_LOGI(TAG, "Initializing Ml307Board");
+    StartModem();
+}
+
+AudioDevice* Ml307Board::CreateAudioDevice() {
+    return new AudioDevice();
+}
+
+Http* Ml307Board::CreateHttp() {
+    return new Ml307Http(modem_);
+}
+
+WebSocket* Ml307Board::CreateWebSocket() {
+    return new WebSocket(new Ml307SslTransport(modem_, 0));
+}
+
+bool Ml307Board::GetNetworkState(std::string& network_name, int& signal_quality, std::string& signal_quality_text) {
+    if (!modem_.network_ready()) {
+        return false;
+    }
+    network_name = modem_.GetCarrierName();
+    signal_quality = modem_.GetCsq();
+    signal_quality_text = csq_to_string(signal_quality);
+    return signal_quality != -1;
+}
+
+std::string Ml307Board::GetJson() {
+    // Set the board type for OTA
+    std::string board_type = BOARD_TYPE;
+    std::string module_name = modem_.GetModuleName();
+    std::string carrier_name = modem_.GetCarrierName();
+    std::string imei = modem_.GetImei();
+    std::string iccid = modem_.GetIccid();
+    int csq = modem_.GetCsq();
+    std::string board_json = std::string("{\"type\":\"" + board_type + "\",");
+    board_json += "\"revision\":\"" + module_name + "\",";
+    board_json += "\"carrier\":\"" + carrier_name + "\",";
+    board_json += "\"csq\":\"" + std::to_string(csq) + "\",";
+    board_json += "\"imei\":\"" + imei + "\",";
+    board_json += "\"iccid\":\"" + iccid + "\"}";
+    return board_json;
+}
--- a/main/Ml307Board.h
+++ b/main/Ml307Board.h
@@ -0,0 +1,24 @@
+#ifndef ML307_BOARD_H
+#define ML307_BOARD_H
+
+#include "Board.h"
+#include <Ml307AtModem.h>
+
+class Ml307Board : public Board {
+protected:
+    Ml307AtModem modem_;
+
+    void StartModem();
+
+public:
+    Ml307Board();
+    virtual void Initialize() override;
+    virtual void StartNetwork() override;
+    virtual AudioDevice* CreateAudioDevice() override;
+    virtual Http* CreateHttp() override;
+    virtual WebSocket* CreateWebSocket() override;
+    virtual bool GetNetworkState(std::string& network_name, int& signal_quality, std::string& signal_quality_text) override;
+    virtual std::string GetJson() override;
+};
+
+#endif // ML307_BOARD_H
--- a/main/SystemInfo.cc
+++ b/main/SystemInfo.cc
@@ -3,7 +3,6 @@
 #include <esp_log.h>
 #include <esp_flash.h>
 #include <esp_mac.h>
-#include <esp_chip_info.h>
 #include <esp_system.h>
 #include <esp_partition.h>
 #include <esp_app_desc.h>
@@ -41,96 +40,6 @@ std::string SystemInfo::GetChipModelName() {
    return std::string(CONFIG_IDF_TARGET);
 }

-std::string SystemInfo::GetJsonString() {
-    /* 
-        {
-            "flash_size": 4194304,
-            "psram_size": 0,
-            "minimum_free_heap_size": 123456,
-            "mac_address": "00:00:00:00:00:00",
-            "chip_model_name": "esp32s3",
-            "chip_info": {
-                "model": 1,
-                "cores": 2,
-                "revision": 0,
-                "features": 0
-            },
-            "application": {
-                "name": "my-app",
-                "version": "1.0.0",
-                "compile_time": "2021-01-01T00:00:00Z"
-                "idf_version": "4.2-dev"
-                "elf_sha256": ""
-            },
-            "partition_table": [
-                "app": {
-                    "label": "app",
-                    "type": 1,
-                    "subtype": 2,
-                    "address": 0x10000,
-                    "size": 0x100000
-                }
-            ],
-            "ota": {
-                "label": "ota_0"
-            }
-        }
-    */
-    std::string json = "{";
-    json += "\"flash_size\":" + std::to_string(GetFlashSize()) + ",";
-    json += "\"minimum_free_heap_size\":" + std::to_string(GetMinimumFreeHeapSize()) + ",";
-    json += "\"mac_address\":\"" + GetMacAddress() + "\",";
-    json += "\"chip_model_name\":\"" + GetChipModelName() + "\",";
-    json += "\"chip_info\":{";
-
-    esp_chip_info_t chip_info;
-    esp_chip_info(&chip_info);
-    json += "\"model\":" + std::to_string(chip_info.model) + ",";
-    json += "\"cores\":" + std::to_string(chip_info.cores) + ",";
-    json += "\"revision\":" + std::to_string(chip_info.revision) + ",";
-    json += "\"features\":" + std::to_string(chip_info.features);
-    json += "},";
-
-    json += "\"application\":{";
-    auto app_desc = esp_app_get_description();
-    json += "\"name\":\"" + std::string(app_desc->project_name) + "\",";
-    json += "\"version\":\"" + std::string(app_desc->version) + "\",";
-    json += "\"compile_time\":\"" + std::string(app_desc->date) + "T" + std::string(app_desc->time) + "Z\",";
-    json += "\"idf_version\":\"" + std::string(app_desc->idf_ver) + "\",";
-
-    char sha256_str[65];
-    for (int i = 0; i < 32; i++) {
-        snprintf(sha256_str + i * 2, sizeof(sha256_str) - i * 2, "%02x", app_desc->app_elf_sha256[i]);
-    }
-    json += "\"elf_sha256\":\"" + std::string(sha256_str) + "\"";
-    json += "},";
-
-    json += "\"partition_table\": [";
-    esp_partition_iterator_t it = esp_partition_find(ESP_PARTITION_TYPE_ANY, ESP_PARTITION_SUBTYPE_ANY, NULL);
-    while (it) {
-        const esp_partition_t *partition = esp_partition_get(it);
-        json += "{";
-        json += "\"label\":\"" + std::string(partition->label) + "\",";
-        json += "\"type\":" + std::to_string(partition->type) + ",";
-        json += "\"subtype\":" + std::to_string(partition->subtype) + ",";
-        json += "\"address\":" + std::to_string(partition->address) + ",";
-        json += "\"size\":" + std::to_string(partition->size);
-        json += "},";
-        it = esp_partition_next(it);
-    }
-    json.pop_back(); // Remove the last comma
-    json += "],";
-
-    json += "\"ota\":{";
-    auto ota_partition = esp_ota_get_running_partition();
-    json += "\"label\":\"" + std::string(ota_partition->label) + "\"";
-    json += "}";
-
-    // Close the JSON object
-    json += "}";
-    return json;
-}
-
 esp_err_t SystemInfo::PrintRealTimeStats(TickType_t xTicksToWait) {
    #define ARRAY_SIZE_OFFSET 5
    TaskStatus_t *start_array = NULL, *end_array = NULL;
--- a/main/SystemInfo.h
+++ b/main/SystemInfo.h
@@ -13,7 +13,6 @@ public:
    static size_t GetFreeHeapSize();
    static std::string GetMacAddress();
    static std::string GetChipModelName();
-    static std::string GetJsonString();
    static esp_err_t PrintRealTimeStats(TickType_t xTicksToWait);
 };

--- a/main/SystemReset.cc
+++ b/main/SystemReset.cc
@@ -41,6 +41,10 @@ void SystemReset::ResetNvsFlash() {
    if (ret != ESP_OK) {
        ESP_LOGE(TAG, "Failed to erase NVS flash");
    }
+    ret = nvs_flash_init();
+    if (ret != ESP_OK) {
+        ESP_LOGE(TAG, "Failed to initialize NVS flash");
+    }
 }

 void SystemReset::ResetToFactory() {
--- a/main/SystemReset.h
+++ b/main/SystemReset.h
@@ -3,11 +3,18 @@

 class SystemReset {
 public:
-    SystemReset();
+    static SystemReset& GetInstance() {
+        static SystemReset instance;
+        return instance;
+    }

    void CheckButtons();

 private:
+    SystemReset(); // 构造函数私有化
+    SystemReset(const SystemReset&) = delete; // 禁用拷贝构造
+    SystemReset& operator=(const SystemReset&) = delete; // 禁用赋值操作
+
    void ResetNvsFlash();
    void ResetToFactory();
    void RestartInSeconds(int seconds);
--- a/main/WakeWordDetect.cc
+++ b/main/WakeWordDetect.cc
@@ -1,5 +1,6 @@
 #include <esp_log.h>
 #include <model_path.h>
+#include <arpa/inet.h>

 #include "WakeWordDetect.h"
 #include "Application.h"
@@ -15,6 +16,24 @@ WakeWordDetect::WakeWordDetect()
      wake_word_opus_() {

    event_group_ = xEventGroupCreate();
+}
+
+WakeWordDetect::~WakeWordDetect() {
+    if (afe_detection_data_ != nullptr) {
+        esp_afe_sr_v1.destroy(afe_detection_data_);
+    }
+
+    if (wake_word_encode_task_stack_ != nullptr) {
+        free(wake_word_encode_task_stack_);
+    }
+
+    vEventGroupDelete(event_group_);
+}
+
+void WakeWordDetect::Initialize(int channels, bool reference) {
+    channels_ = channels;
+    reference_ = reference;
+    int ref_num = reference_ ? 1 : 0;

    srmodel_list_t *models = esp_srmodel_init("model");
    for (int i = 0; i < models->num; i++) {
@@ -25,7 +44,7 @@ WakeWordDetect::WakeWordDetect()
    }

    afe_config_t afe_config = {
-        .aec_init = false,
+        .aec_init = reference_,
        .se_init = true,
        .vad_init = true,
        .wakenet_init = true,
@@ -37,17 +56,17 @@ WakeWordDetect::WakeWordDetect()
        .wakenet_model_name_2 = NULL,
        .wakenet_mode = DET_MODE_90,
        .afe_mode = SR_MODE_HIGH_PERF,
-        .afe_perferred_core = 0,
-        .afe_perferred_priority = 5,
+        .afe_perferred_core = 1,
+        .afe_perferred_priority = 1,
        .afe_ringbuf_size = 50,
        .memory_alloc_mode = AFE_MEMORY_ALLOC_MORE_PSRAM,
        .afe_linear_gain = 1.0,
        .agc_mode = AFE_MN_PEAK_AGC_MODE_2,
        .pcm_config = {
-            .total_ch_num = 1,
-            .mic_num = 1,
-            .ref_num = 0,
-            .sample_rate = CONFIG_AUDIO_INPUT_SAMPLE_RATE
+            .total_ch_num = channels_,
+            .mic_num = channels_ - ref_num,
+            .ref_num = ref_num,
+            .sample_rate = 16000
        },
        .debug_init = false,
        .debug_hook = {{ AFE_DEBUG_HOOK_MASE_TASK_IN, NULL }, { AFE_DEBUG_HOOK_FETCH_TASK_IN, NULL }},
@@ -62,19 +81,7 @@ WakeWordDetect::WakeWordDetect()
        auto this_ = (WakeWordDetect*)arg;
        this_->AudioDetectionTask();
        vTaskDelete(NULL);
-    }, "audio_detection", 4096 * 2, this, 5, NULL);
-}
-
-WakeWordDetect::~WakeWordDetect() {
-    if (afe_detection_data_ != nullptr) {
-        esp_afe_sr_v1.destroy(afe_detection_data_);
-    }
-
-    if (wake_word_encode_task_stack_ != nullptr) {
-        free(wake_word_encode_task_stack_);
-    }
-
-    vEventGroupDelete(event_group_);
+    }, "audio_detection", 4096 * 2, this, 1, NULL);
 }

 void WakeWordDetect::OnWakeWordDetected(std::function<void()> callback) {
@@ -97,10 +104,10 @@ bool WakeWordDetect::IsDetectionRunning() {
    return xEventGroupGetBits(event_group_) & DETECTION_RUNNING_EVENT;
 }

-void WakeWordDetect::Feed(const int16_t* data, int size) {
-    input_buffer_.insert(input_buffer_.end(), data, data + size);
+void WakeWordDetect::Feed(std::vector<int16_t>& data) {
+    input_buffer_.insert(input_buffer_.end(), data.begin(), data.end());

-    auto chunk_size = esp_afe_sr_v1.get_feed_chunksize(afe_detection_data_);
+    auto chunk_size = esp_afe_sr_v1.get_feed_chunksize(afe_detection_data_) * channels_;
    while (input_buffer_.size() >= chunk_size) {
        esp_afe_sr_v1.feed(afe_detection_data_, input_buffer_.data());
        input_buffer_.erase(input_buffer_.begin(), input_buffer_.begin() + chunk_size);
@@ -166,21 +173,19 @@ void WakeWordDetect::EncodeWakeWordData() {
        auto start_time = esp_timer_get_time();
        // encode detect packets
        OpusEncoder* encoder = new OpusEncoder();
-        encoder->Configure(CONFIG_AUDIO_INPUT_SAMPLE_RATE, 1, 60);
+        encoder->Configure(16000, 1, 60);
        encoder->SetComplexity(0);
        this_->wake_word_opus_.resize(4096 * 4);
        size_t offset = 0;

        for (auto& pcm: this_->wake_word_pcm_) {
            encoder->Encode(pcm, [this_, &offset](const uint8_t* opus, size_t opus_size) {
-                size_t protocol_size = sizeof(BinaryProtocol) + opus_size;
+                size_t protocol_size = sizeof(BinaryProtocol3) + opus_size;
                if (offset + protocol_size < this_->wake_word_opus_.size()) {
-                    auto protocol = (BinaryProtocol*)(&this_->wake_word_opus_[offset]);
-                    protocol->version = htons(PROTOCOL_VERSION);
-                    protocol->type = htons(0);
+                    auto protocol = (BinaryProtocol3*)(&this_->wake_word_opus_[offset]);
+                    protocol->type = 0;
                    protocol->reserved = 0;
-                    protocol->timestamp = 0;
-                    protocol->payload_size = htonl(opus_size);
+                    protocol->payload_size = htons(opus_size);
                    memcpy(protocol->payload, opus, opus_size);
                    offset += protocol_size;
                }
--- a/main/WakeWordDetect.h
+++ b/main/WakeWordDetect.h
@@ -19,7 +19,8 @@ public:
    WakeWordDetect();
    ~WakeWordDetect();

-    void Feed(const int16_t* data, int size);
+    void Initialize(int channels, bool reference);
+    void Feed(std::vector<int16_t>& data);
    void OnWakeWordDetected(std::function<void()> callback);
    void OnVadStateChange(std::function<void(bool speaking)> callback);
    void StartDetection();
@@ -36,6 +37,8 @@ private:
    std::function<void()> wake_word_detected_callback_;
    std::function<void(bool speaking)> vad_state_change_callback_;
    bool is_speaking_ = false;
+    int channels_;
+    bool reference_;

    TaskHandle_t wake_word_encode_task_ = nullptr;
    StaticTask_t wake_word_encode_task_buffer_;
--- a/main/WifiBoard.cc
+++ b/main/WifiBoard.cc
@@ -0,0 +1,103 @@
+#include "WifiBoard.h"
+#include "Application.h"
+#include "WifiStation.h"
+#include "WifiConfigurationAp.h"
+#include "SystemInfo.h"
+#include "BuiltinLed.h"
+
+#include <freertos/FreeRTOS.h>
+#include <freertos/task.h>
+#include <EspHttp.h>
+#include <TcpTransport.h>
+#include <TlsTransport.h>
+#include <WebSocket.h>
+#include <esp_log.h>
+
+static const char *TAG = "WifiBoard";
+
+static std::string rssi_to_string(int rssi) {
+    if (rssi >= -55) {
+        return "Very good";
+    } else if (rssi >= -65) {
+        return "Good";
+    } else if (rssi >= -75) {
+        return "Fair";
+    } else if (rssi >= -85) {
+        return "Poor";
+    } else {
+        return "No network";
+    }
+}
+
+void WifiBoard::StartNetwork() {
+    auto& application = Application::GetInstance();
+    auto& display = application.GetDisplay();
+    auto& builtin_led = BuiltinLed::GetInstance();
+
+    // Try to connect to WiFi, if failed, launch the WiFi configuration AP
+    auto& wifi_station = WifiStation::GetInstance();
+    display.SetText(std::string("Connect to WiFi\n") + wifi_station.GetSsid());
+    wifi_station.Start();
+    if (!wifi_station.IsConnected()) {
+        application.Alert("Info", "Configuring WiFi");
+        builtin_led.SetBlue();
+        builtin_led.Blink(1000, 500);
+        auto& wifi_ap = WifiConfigurationAp::GetInstance();
+        wifi_ap.SetSsidPrefix("Xiaozhi");
+        wifi_ap.Start();
+        // Wait forever until reset after configuration
+        while (true) {
+            vTaskDelay(pdMS_TO_TICKS(1000));
+        }
+    }
+}
+
+void WifiBoard::Initialize() {
+    ESP_LOGI(TAG, "Initializing WifiBoard");
+}
+
+Http* WifiBoard::CreateHttp() {
+    return new EspHttp();
+}
+
+WebSocket* WifiBoard::CreateWebSocket() {
+    std::string url = CONFIG_WEBSOCKET_URL;
+    if (url.find("wss://") == 0) {
+            return new WebSocket(new TlsTransport());
+        } else {
+            return new WebSocket(new TcpTransport());
+        }
+}
+
+bool WifiBoard::GetNetworkState(std::string& network_name, int& signal_quality, std::string& signal_quality_text) {
+    if (wifi_config_mode_) {
+        auto& wifi_ap = WifiConfigurationAp::GetInstance();
+        network_name = wifi_ap.GetSsid();
+        signal_quality = -99;
+        signal_quality_text = wifi_ap.GetWebServerUrl();
+        return true;
+    }
+    auto& wifi_station = WifiStation::GetInstance();
+    if (!wifi_station.IsConnected()) {
+        return false;
+    }
+    network_name = wifi_station.GetSsid();
+    signal_quality = wifi_station.GetRssi();
+    signal_quality_text = rssi_to_string(signal_quality);
+    return signal_quality != -1;
+}
+
+std::string WifiBoard::GetJson() {
+    // Set the board type for OTA
+    auto& wifi_station = WifiStation::GetInstance();
+    std::string board_type = BOARD_TYPE;
+    std::string board_json = std::string("{\"type\":\"" + board_type + "\",");
+    if (!wifi_config_mode_) {
+        board_json += "\"ssid\":\"" + wifi_station.GetSsid() + "\",";
+        board_json += "\"rssi\":" + std::to_string(wifi_station.GetRssi()) + ",";
+        board_json += "\"channel\":" + std::to_string(wifi_station.GetChannel()) + ",";
+        board_json += "\"ip\":\"" + wifi_station.GetIpAddress() + "\",";
+    }
+    board_json += "\"mac\":\"" + SystemInfo::GetMacAddress() + "\"}";
+    return board_json;
+}
--- a/main/WifiBoard.h
+++ b/main/WifiBoard.h
@@ -0,0 +1,19 @@
+#ifndef WIFI_BOARD_H
+#define WIFI_BOARD_H
+
+#include "Board.h"
+
+class WifiBoard : public Board {
+protected:
+    bool wifi_config_mode_ = false;
+
+public:
+    virtual void Initialize() override;
+    virtual void StartNetwork() override;
+    virtual Http* CreateHttp() override;
+    virtual WebSocket* CreateWebSocket() override;
+    virtual bool GetNetworkState(std::string& network_name, int& signal_quality, std::string& signal_quality_text) override;
+    virtual std::string GetJson() override;
+};
+
+#endif // WIFI_BOARD_H
--- a/main/assets/err_pin.p3
+++ b/main/assets/err_pin.p3
--- a/main/assets/err_reg.p3
+++ b/main/assets/err_reg.p3
--- a/main/assets/err_wificonfig.p3
+++ b/main/assets/err_wificonfig.p3
--- a/main/boards/bread-compact-ml307/CompactMl307Board.cc
+++ b/main/boards/bread-compact-ml307/CompactMl307Board.cc
@@ -0,0 +1,22 @@
+#include "Ml307Board.h"
+#include "SystemReset.h"
+#include <esp_log.h>
+
+#define TAG "CompactMl307Board"
+
+class CompactMl307Board : public Ml307Board {
+public:
+    virtual void Initialize() override {
+        ESP_LOGI(TAG, "Initializing CompactMl307Board");
+        // Check if the reset button is pressed
+        SystemReset::GetInstance().CheckButtons();
+
+        Ml307Board::Initialize();
+    }
+
+    virtual AudioDevice* CreateAudioDevice() override {
+        return new AudioDevice();
+    }
+};
+
+DECLARE_BOARD(CompactMl307Board);
--- a/main/boards/bread-compact-ml307/config.h
+++ b/main/boards/bread-compact-ml307/config.h
@@ -0,0 +1,46 @@
+#ifndef _BOARD_CONFIG_H_
+#define _BOARD_CONFIG_H_
+
+#include <driver/gpio.h>
+
+#define AUDIO_INPUT_SAMPLE_RATE  16000
+#define AUDIO_OUTPUT_SAMPLE_RATE 24000
+
+#define AUDIO_I2S_METHOD_SIMPLEX
+
+#ifdef AUDIO_I2S_METHOD_SIMPLEX
+
+#define AUDIO_I2S_MIC_GPIO_WS   GPIO_NUM_4
+#define AUDIO_I2S_MIC_GPIO_SCK  GPIO_NUM_5
+#define AUDIO_I2S_MIC_GPIO_DIN  GPIO_NUM_6
+#define AUDIO_I2S_SPK_GPIO_DOUT GPIO_NUM_7
+#define AUDIO_I2S_SPK_GPIO_BCLK GPIO_NUM_15
+#define AUDIO_I2S_SPK_GPIO_LRCK GPIO_NUM_16
+
+#else
+
+#define AUDIO_I2S_GPIO_LRCK GPIO_NUM_4
+#define AUDIO_I2S_GPIO_BCLK GPIO_NUM_5
+#define AUDIO_I2S_GPIO_DIN  GPIO_NUM_6
+#define AUDIO_I2S_GPIO_DOUT GPIO_NUM_7
+
+#endif
+
+#define BUILTIN_LED_GPIO        GPIO_NUM_48
+#define BOOT_BUTTON_GPIO        GPIO_NUM_0
+#define VOLUME_UP_BUTTON_GPIO   GPIO_NUM_40
+#define VOLUME_DOWN_BUTTON_GPIO GPIO_NUM_39
+
+#define DISPLAY_SDA_PIN GPIO_NUM_41
+#define DISPLAY_SCL_PIN GPIO_NUM_42
+#define DISPLAY_WIDTH   128
+#define DISPLAY_HEIGHT  32
+#define DISPLAY_MIRROR_X true
+#define DISPLAY_MIRROR_Y true
+
+
+#define ML307_RX_PIN GPIO_NUM_11
+#define ML307_TX_PIN GPIO_NUM_12
+
+
+#endif // _BOARD_CONFIG_H_
--- a/main/boards/bread-compact-wifi/CompactWifiBoard.cc
+++ b/main/boards/bread-compact-wifi/CompactWifiBoard.cc
@@ -0,0 +1,22 @@
+#include "WifiBoard.h"
+#include "SystemReset.h"
+#include <esp_log.h>
+
+#define TAG "CompactWifiBoard"
+
+class CompactWifiBoard : public WifiBoard {
+public:
+    virtual void Initialize() override {
+        ESP_LOGI(TAG, "Initializing CompactWifiBoard");
+        // Check if the reset button is pressed
+        SystemReset::GetInstance().CheckButtons();
+
+        WifiBoard::Initialize();
+    }
+
+    virtual AudioDevice* CreateAudioDevice() override {
+        return new AudioDevice();
+    }
+};
+
+DECLARE_BOARD(CompactWifiBoard);
--- a/main/boards/bread-compact-wifi/config.h
+++ b/main/boards/bread-compact-wifi/config.h
@@ -0,0 +1,42 @@
+#ifndef _BOARD_CONFIG_H_
+#define _BOARD_CONFIG_H_
+
+#include <driver/gpio.h>
+
+#define AUDIO_INPUT_SAMPLE_RATE  16000
+#define AUDIO_OUTPUT_SAMPLE_RATE 24000
+
+#define AUDIO_I2S_METHOD_SIMPLEX
+
+#ifdef AUDIO_I2S_METHOD_SIMPLEX
+
+#define AUDIO_I2S_MIC_GPIO_WS   GPIO_NUM_4
+#define AUDIO_I2S_MIC_GPIO_SCK  GPIO_NUM_5
+#define AUDIO_I2S_MIC_GPIO_DIN  GPIO_NUM_6
+#define AUDIO_I2S_SPK_GPIO_DOUT GPIO_NUM_7
+#define AUDIO_I2S_SPK_GPIO_BCLK GPIO_NUM_15
+#define AUDIO_I2S_SPK_GPIO_LRCK GPIO_NUM_16
+
+#else
+
+#define AUDIO_I2S_GPIO_LRCK GPIO_NUM_4
+#define AUDIO_I2S_GPIO_BCLK GPIO_NUM_5
+#define AUDIO_I2S_GPIO_DIN  GPIO_NUM_6
+#define AUDIO_I2S_GPIO_DOUT GPIO_NUM_7
+
+#endif
+
+
+#define BUILTIN_LED_GPIO        GPIO_NUM_48
+#define BOOT_BUTTON_GPIO        GPIO_NUM_0
+#define VOLUME_UP_BUTTON_GPIO   GPIO_NUM_40
+#define VOLUME_DOWN_BUTTON_GPIO GPIO_NUM_39
+
+#define DISPLAY_SDA_PIN GPIO_NUM_41
+#define DISPLAY_SCL_PIN GPIO_NUM_42
+#define DISPLAY_WIDTH   128
+#define DISPLAY_HEIGHT  32
+#define DISPLAY_MIRROR_X true
+#define DISPLAY_MIRROR_Y true
+
+#endif // _BOARD_CONFIG_H_
--- a/main/boards/esp-box-3/EspBox3Board.cc
+++ b/main/boards/esp-box-3/EspBox3Board.cc
@@ -0,0 +1,20 @@
+#include "WifiBoard.h"
+#include "BoxAudioDevice.h"
+
+#include <esp_log.h>
+
+#define TAG "EspBox3Board"
+
+class EspBox3Board : public WifiBoard {
+public:
+    virtual void Initialize() override {
+        ESP_LOGI(TAG, "Initializing EspBox3Board");
+        WifiBoard::Initialize();
+    }
+
+    virtual AudioDevice* CreateAudioDevice() override {
+        return new BoxAudioDevice();
+    }
+};
+
+DECLARE_BOARD(EspBox3Board);
--- a/main/boards/esp-box-3/config.h
+++ b/main/boards/esp-box-3/config.h
@@ -0,0 +1,36 @@
+#ifndef _BOARD_CONFIG_H_
+#define _BOARD_CONFIG_H_
+
+#include <driver/gpio.h>
+
+#define AUDIO_INPUT_SAMPLE_RATE  16000
+#define AUDIO_OUTPUT_SAMPLE_RATE 16000
+
+#define AUDIO_INPUT_REFERENCE    true
+
+#define AUDIO_I2S_GPIO_MCLK GPIO_NUM_2
+#define AUDIO_I2S_GPIO_LRCK GPIO_NUM_45
+#define AUDIO_I2S_GPIO_BCLK GPIO_NUM_17
+#define AUDIO_I2S_GPIO_DIN  GPIO_NUM_16
+#define AUDIO_I2S_GPIO_DOUT GPIO_NUM_15
+
+#define AUDIO_CODEC_PA_PIN       GPIO_NUM_46
+#define AUDIO_CODEC_I2C_SDA_PIN  GPIO_NUM_8
+#define AUDIO_CODEC_I2C_SCL_PIN  GPIO_NUM_18
+#define AUDIO_CODEC_ES8311_ADDR  ES8311_CODEC_DEFAULT_ADDR
+#define AUDIO_CODEC_ES7210_ADDR  ES7210_CODEC_DEFAULT_ADDR
+
+#define BUILTIN_LED_GPIO        GPIO_NUM_NC
+#define BOOT_BUTTON_GPIO        GPIO_NUM_0
+#define VOLUME_UP_BUTTON_GPIO   GPIO_NUM_NC
+#define VOLUME_DOWN_BUTTON_GPIO GPIO_NUM_NC
+
+#define DISPLAY_SDA_PIN GPIO_NUM_NC
+#define DISPLAY_SCL_PIN GPIO_NUM_NC
+#define DISPLAY_WIDTH   128
+#define DISPLAY_HEIGHT  64
+#define DISPLAY_MIRROR_X true
+#define DISPLAY_MIRROR_Y true
+
+
+#endif // _BOARD_CONFIG_H_
--- a/main/boards/kevin-box-0/KevinBoxBoard.cc
+++ b/main/boards/kevin-box-0/KevinBoxBoard.cc
@@ -0,0 +1,84 @@
+#include "Ml307Board.h"
+#include "BoxAudioDevice.h"
+
+#include <esp_log.h>
+#include <esp_spiffs.h>
+#include <driver/gpio.h>
+#include <esp_adc/adc_oneshot.h>
+#include <esp_adc/adc_cali.h>
+#include <esp_adc/adc_cali_scheme.h>
+
+static const char *TAG = "KevinBoxBoard";
+
+class KevinBoxBoard : public Ml307Board {
+private:
+    adc_oneshot_unit_handle_t adc1_handle_;
+    adc_cali_handle_t adc1_cali_handle_;
+
+    void MountStorage() {
+        // Mount the storage partition
+        esp_vfs_spiffs_conf_t conf = {
+            .base_path = "/storage",
+            .partition_label = "storage",
+            .max_files = 5,
+            .format_if_mount_failed = true,
+        };
+        esp_vfs_spiffs_register(&conf);
+    }
+
+    void Enable4GModule() {
+        // Make GPIO15 HIGH to enable the 4G module
+        gpio_config_t ml307_enable_config = {
+            .pin_bit_mask = (1ULL << 15),
+            .mode = GPIO_MODE_OUTPUT,
+            .pull_up_en = GPIO_PULLUP_DISABLE,
+            .pull_down_en = GPIO_PULLDOWN_DISABLE,
+            .intr_type = GPIO_INTR_DISABLE,
+        };
+        gpio_config(&ml307_enable_config);
+        gpio_set_level(GPIO_NUM_15, 1);
+    }
+
+    virtual void InitializeADC() {
+        adc_oneshot_unit_init_cfg_t init_config1 = {};
+        init_config1.unit_id = ADC_UNIT_1;
+        ESP_ERROR_CHECK(adc_oneshot_new_unit(&init_config1, &adc1_handle_));
+
+        //-------------ADC1 Config---------------//
+        adc_oneshot_chan_cfg_t config = {
+            .atten = ADC_ATTEN_DB_12,
+            .bitwidth = ADC_BITWIDTH_DEFAULT,
+        };
+        ESP_ERROR_CHECK(adc_oneshot_config_channel(adc1_handle_, ADC_CHANNEL_0, &config));
+
+        adc_cali_curve_fitting_config_t cali_config = {
+            .unit_id = ADC_UNIT_1,
+            .chan = ADC_CHANNEL_0,
+            .atten = ADC_ATTEN_DB_12,
+            .bitwidth = ADC_BITWIDTH_DEFAULT,
+        };
+        ESP_ERROR_CHECK(adc_cali_create_scheme_curve_fitting(&cali_config, &adc1_cali_handle_));
+    }
+public:
+    virtual void Initialize() override {
+        ESP_LOGI(TAG, "Initializing KevinBoxBoard");
+        InitializeADC();
+        MountStorage();
+        Enable4GModule();
+        Ml307Board::Initialize();
+    }
+
+    virtual AudioDevice* CreateAudioDevice() override {
+        return new BoxAudioDevice();
+    }
+
+    virtual bool GetBatteryVoltage(int &voltage, bool& charging) override {
+        int adc_reading;
+        ESP_ERROR_CHECK(adc_oneshot_read(adc1_handle_, ADC_CHANNEL_0, &adc_reading));
+        ESP_ERROR_CHECK(adc_cali_raw_to_voltage(adc1_cali_handle_, adc_reading, &voltage));
+        charging = false;
+        return true;
+    }
+};
+
+DECLARE_BOARD(KevinBoxBoard);
--- a/main/boards/kevin-box-0/config.h
+++ b/main/boards/kevin-box-0/config.h
@@ -0,0 +1,39 @@
+#ifndef _BOARD_CONFIG_H_
+#define _BOARD_CONFIG_H_
+
+#include <driver/gpio.h>
+
+#define AUDIO_INPUT_SAMPLE_RATE  24000
+#define AUDIO_OUTPUT_SAMPLE_RATE 24000
+
+#define AUDIO_INPUT_REFERENCE    true
+
+#define AUDIO_I2S_GPIO_MCLK GPIO_NUM_0
+#define AUDIO_I2S_GPIO_LRCK GPIO_NUM_47
+#define AUDIO_I2S_GPIO_BCLK GPIO_NUM_48
+#define AUDIO_I2S_GPIO_DIN  GPIO_NUM_45
+#define AUDIO_I2S_GPIO_DOUT GPIO_NUM_21
+
+#define AUDIO_CODEC_PA_PIN       GPIO_NUM_40
+#define AUDIO_CODEC_I2C_SDA_PIN  GPIO_NUM_39
+#define AUDIO_CODEC_I2C_SCL_PIN  GPIO_NUM_38
+#define AUDIO_CODEC_ES8311_ADDR  ES8311_CODEC_DEFAULT_ADDR
+#define AUDIO_CODEC_ES7210_ADDR  ES7210_CODEC_DEFAULT_ADDR
+
+#define BUILTIN_LED_GPIO        GPIO_NUM_8
+#define BOOT_BUTTON_GPIO        GPIO_NUM_0
+#define VOLUME_UP_BUTTON_GPIO   GPIO_NUM_6
+#define VOLUME_DOWN_BUTTON_GPIO GPIO_NUM_7
+
+#define DISPLAY_SDA_PIN GPIO_NUM_4
+#define DISPLAY_SCL_PIN GPIO_NUM_5
+#define DISPLAY_WIDTH   128
+#define DISPLAY_HEIGHT  64
+#define DISPLAY_MIRROR_X true
+#define DISPLAY_MIRROR_Y true
+
+#define ML307_RX_PIN GPIO_NUM_17
+#define ML307_TX_PIN GPIO_NUM_16
+
+
+#endif // _BOARD_CONFIG_H_
--- a/main/boards/kevin-box-1/KevinBoxBoard.cc
+++ b/main/boards/kevin-box-1/KevinBoxBoard.cc
@@ -0,0 +1,93 @@
+#include "Ml307Board.h"
+#include "BoxAudioDevice.h"
+
+#include <esp_log.h>
+#include <esp_spiffs.h>
+#include <driver/gpio.h>
+#include <esp_adc/adc_oneshot.h>
+#include <esp_adc/adc_cali.h>
+#include <esp_adc/adc_cali_scheme.h>
+
+static const char *TAG = "KevinBoxBoard";
+
+class KevinBoxBoard : public Ml307Board {
+private:
+    adc_oneshot_unit_handle_t adc1_handle_;
+    adc_cali_handle_t adc1_cali_handle_;
+
+    void MountStorage() {
+        // Mount the storage partition
+        esp_vfs_spiffs_conf_t conf = {
+            .base_path = "/storage",
+            .partition_label = "storage",
+            .max_files = 5,
+            .format_if_mount_failed = true,
+        };
+        esp_vfs_spiffs_register(&conf);
+    }
+
+    void Enable4GModule() {
+        // Make GPIO15 HIGH to enable the 4G module
+        gpio_config_t ml307_enable_config = {
+            .pin_bit_mask = (1ULL << 15),
+            .mode = GPIO_MODE_OUTPUT,
+            .pull_up_en = GPIO_PULLUP_DISABLE,
+            .pull_down_en = GPIO_PULLDOWN_DISABLE,
+            .intr_type = GPIO_INTR_DISABLE,
+        };
+        gpio_config(&ml307_enable_config);
+        gpio_set_level(GPIO_NUM_15, 1);
+    }
+
+    virtual void InitializeADC() {
+        adc_oneshot_unit_init_cfg_t init_config1 = {};
+        init_config1.unit_id = ADC_UNIT_1;
+        ESP_ERROR_CHECK(adc_oneshot_new_unit(&init_config1, &adc1_handle_));
+
+        //-------------ADC1 Config---------------//
+        adc_oneshot_chan_cfg_t config = {
+            .atten = ADC_ATTEN_DB_12,
+            .bitwidth = ADC_BITWIDTH_DEFAULT,
+        };
+        ESP_ERROR_CHECK(adc_oneshot_config_channel(adc1_handle_, ADC_CHANNEL_0, &config));
+
+        adc_cali_curve_fitting_config_t cali_config = {
+            .unit_id = ADC_UNIT_1,
+            .chan = ADC_CHANNEL_0,
+            .atten = ADC_ATTEN_DB_12,
+            .bitwidth = ADC_BITWIDTH_DEFAULT,
+        };
+        ESP_ERROR_CHECK(adc_cali_create_scheme_curve_fitting(&cali_config, &adc1_cali_handle_));
+    }
+public:
+    virtual void Initialize() override {
+        ESP_LOGI(TAG, "Initializing KevinBoxBoard");
+        InitializeADC();
+        MountStorage();
+        Enable4GModule();
+
+        gpio_config_t charging_io = {
+            .pin_bit_mask = (1ULL << 2),
+            .mode = GPIO_MODE_INPUT,
+            .pull_up_en = GPIO_PULLUP_ENABLE,
+            .pull_down_en = GPIO_PULLDOWN_DISABLE,
+            .intr_type = GPIO_INTR_DISABLE,
+        };
+        gpio_config(&charging_io);
+
+        Ml307Board::Initialize();
+    }
+
+    virtual AudioDevice* CreateAudioDevice() override {
+        return new BoxAudioDevice();
+    }
+
+    virtual bool GetBatteryVoltage(int &voltage, bool& charging) override {
+        ESP_ERROR_CHECK(adc_oneshot_get_calibrated_result(adc1_handle_, adc1_cali_handle_, ADC_CHANNEL_0, &voltage));
+        charging = gpio_get_level(GPIO_NUM_2) == 0;
+        ESP_LOGI(TAG, "Battery voltage: %d, Charging: %d", voltage, charging);
+        return true;
+    }
+};
+
+DECLARE_BOARD(KevinBoxBoard);
--- a/main/boards/kevin-box-1/config.h
+++ b/main/boards/kevin-box-1/config.h
@@ -0,0 +1,39 @@
+#ifndef _BOARD_CONFIG_H_
+#define _BOARD_CONFIG_H_
+
+#include <driver/gpio.h>
+
+#define AUDIO_INPUT_SAMPLE_RATE  24000
+#define AUDIO_OUTPUT_SAMPLE_RATE 24000
+
+#define AUDIO_INPUT_REFERENCE    true
+
+#define AUDIO_I2S_GPIO_MCLK GPIO_NUM_42
+#define AUDIO_I2S_GPIO_LRCK GPIO_NUM_47
+#define AUDIO_I2S_GPIO_BCLK GPIO_NUM_48
+#define AUDIO_I2S_GPIO_DIN  GPIO_NUM_45
+#define AUDIO_I2S_GPIO_DOUT GPIO_NUM_21
+
+#define AUDIO_CODEC_PA_PIN       GPIO_NUM_17
+#define AUDIO_CODEC_I2C_SDA_PIN  GPIO_NUM_39
+#define AUDIO_CODEC_I2C_SCL_PIN  GPIO_NUM_38
+#define AUDIO_CODEC_ES8311_ADDR  ES8311_CODEC_DEFAULT_ADDR
+#define AUDIO_CODEC_ES7210_ADDR  ES7210_CODEC_DEFAULT_ADDR
+
+#define BUILTIN_LED_GPIO        GPIO_NUM_8
+#define BOOT_BUTTON_GPIO        GPIO_NUM_0
+#define VOLUME_UP_BUTTON_GPIO   GPIO_NUM_6
+#define VOLUME_DOWN_BUTTON_GPIO GPIO_NUM_7
+
+#define DISPLAY_SDA_PIN GPIO_NUM_4
+#define DISPLAY_SCL_PIN GPIO_NUM_5
+#define DISPLAY_WIDTH   128
+#define DISPLAY_HEIGHT  64
+#define DISPLAY_MIRROR_X false
+#define DISPLAY_MIRROR_Y false
+
+#define ML307_RX_PIN GPIO_NUM_20
+#define ML307_TX_PIN GPIO_NUM_19
+
+
+#endif // _BOARD_CONFIG_H_
--- a/main/boards/lichuang-dev/LiChuangDevBoard.cc
+++ b/main/boards/lichuang-dev/LiChuangDevBoard.cc
@@ -0,0 +1,20 @@
+#include "WifiBoard.h"
+#include "BoxAudioDevice.h"
+
+#include <esp_log.h>
+
+#define TAG "LiChuangDevBoard"
+
+class LiChuangDevBoard : public WifiBoard {
+public:
+    virtual void Initialize() override {
+        ESP_LOGI(TAG, "Initializing LiChuangDevBoard");
+        WifiBoard::Initialize();
+    }
+
+    virtual AudioDevice* CreateAudioDevice() override {
+        return new BoxAudioDevice();
+    }
+};
+
+DECLARE_BOARD(LiChuangDevBoard);
--- a/main/boards/lichuang-dev/config.h
+++ b/main/boards/lichuang-dev/config.h
@@ -0,0 +1,37 @@
+#ifndef _BOARD_CONFIG_H_
+#define _BOARD_CONFIG_H_
+
+#include <driver/gpio.h>
+
+#define AUDIO_INPUT_SAMPLE_RATE  24000
+#define AUDIO_OUTPUT_SAMPLE_RATE 24000
+
+#define AUDIO_INPUT_REFERENCE    true
+
+#define AUDIO_I2S_GPIO_MCLK GPIO_NUM_38
+#define AUDIO_I2S_GPIO_LRCK GPIO_NUM_13
+#define AUDIO_I2S_GPIO_BCLK GPIO_NUM_14
+#define AUDIO_I2S_GPIO_DIN  GPIO_NUM_12
+#define AUDIO_I2S_GPIO_DOUT GPIO_NUM_45
+
+#define AUDIO_CODEC_USE_PCA9557
+#define AUDIO_CODEC_PA_PIN       GPIO_NUM_40
+#define AUDIO_CODEC_I2C_SDA_PIN  GPIO_NUM_1
+#define AUDIO_CODEC_I2C_SCL_PIN  GPIO_NUM_2
+#define AUDIO_CODEC_ES8311_ADDR  ES8311_CODEC_DEFAULT_ADDR
+#define AUDIO_CODEC_ES7210_ADDR  0x82
+
+#define BUILTIN_LED_GPIO        GPIO_NUM_48
+#define BOOT_BUTTON_GPIO        GPIO_NUM_0
+#define VOLUME_UP_BUTTON_GPIO   GPIO_NUM_NC
+#define VOLUME_DOWN_BUTTON_GPIO GPIO_NUM_NC
+
+#define DISPLAY_SDA_PIN GPIO_NUM_NC
+#define DISPLAY_SCL_PIN GPIO_NUM_NC
+#define DISPLAY_WIDTH   128
+#define DISPLAY_HEIGHT  64
+#define DISPLAY_MIRROR_X false
+#define DISPLAY_MIRROR_Y false
+
+
+#endif // _BOARD_CONFIG_H_
--- a/main/idf_component.yml
+++ b/main/idf_component.yml
@@ -1,13 +1,14 @@
 ## IDF Component Manager Manifest File
 dependencies:
-  78/esp-builtin-led: "^1.0.2"
-  78/esp-wifi-connect: "^1.1.0"
-  78/esp-opus-encoder: "^1.0.2"
-  78/esp-ml307: "^1.2.1"
+  78/esp-wifi-connect: "~1.2.0"
+  78/esp-opus-encoder: "~1.0.2"
+  78/esp-ml307: "~1.3.0"
+  espressif/led_strip: "^2.4.1"
+  espressif/esp_codec_dev: "^1.3.1"
  espressif/esp-sr: "^1.9.0"
  espressif/button: "^3.3.1"
-  lvgl/lvgl: "^8.4.0"
-  esp_lvgl_port: "^1.4.0"
+  lvgl/lvgl: "~8.4.0"
+  esp_lvgl_port: "~2.4.1"
  ## Required IDF version
  idf:
    version: ">=5.3"
--- a/main/main.cc
+++ b/main/main.cc
@@ -7,16 +7,11 @@

 #include "Application.h"
 #include "SystemInfo.h"
-#include "SystemReset.h"

 #define TAG "main"

 extern "C" void app_main(void)
 {
-    // Check if the reset button is pressed
-    SystemReset system_reset;
-    system_reset.CheckButtons();
-
    // Initialize the default event loop
    ESP_ERROR_CHECK(esp_event_loop_create_default());

--- a/pack.py
+++ b/pack.py
@@ -1,68 +0,0 @@
-#! /usr/bin/env python3
-
-import csv
-import os
-
-# 例如：1000, 0x1000, 1M
-def read_value(text):
-    text = text.strip()
-    if text.endswith('K'):
-        return int(text[:-1]) * 1024
-    elif text.endswith('M'):
-        return int(text[:-1]) * 1024 * 1024
-    else:
-        if text.startswith('0x'):
-            return int(text, 16)
-        else:
-            return int(text)
-
-
-def write_bin(image_data, offset, file_path, max_size=None):
-    # Read file_path and write to image_data
-    with open(file_path, 'rb') as f:
-        data = f.read()
-        if max_size is not None:
-            assert len(data) <= max_size, f"Data from {file_path} is too large"
-        image_data[offset:offset+len(data)] = data
-        print(f"Write {os.path.basename(file_path)} to 0x{offset:08X} with size 0x{len(data):08X}")
-
-
-'''
-根据 partitions.csv 文件，把 bin 文件打包成一个 4MB 的 image 文件，方便烧录
-'''
-def pack_firmware_image():
-    # Create a 4MB image filled with 0xFF
-    image_size = 4 * 1024 * 1024
-    image_data = bytearray([0xFF] * image_size)
-
-    build_dir = os.path.join(os.path.dirname(__file__), 'build')
-    write_bin(image_data, 0, os.path.join(build_dir, 'bootloader', 'bootloader.bin'))
-    write_bin(image_data, 0x8000, os.path.join(build_dir, 'partition_table', 'partition-table.bin'))
-
-    # 读取 partitions.csv 文件
-    with open('partitions.csv', 'r') as f:
-        reader = csv.reader(f)
-        for row in reader:
-            if row[0] == 'model':   
-                file_path = os.path.join(build_dir, 'srmodels', 'srmodels.bin')
-            elif row[0] == 'factory':
-                file_path = os.path.join(build_dir, 'xiaozhi.bin')
-            else:
-                continue
-
-            offset = read_value(row[3])
-            size = read_value(row[4])
-            write_bin(image_data, offset, file_path, size)
-
-    # 写入 image 文件
-    output_path = os.path.join(build_dir, 'xiaozhi.img')
-    with open(output_path, 'wb') as f:
-        f.write(image_data)
-    print(f"Image file {output_path} created with size 0x{len(image_data):08X}")
-
-    # Compress image with zip without directory
-    os.system(f"zip -j {output_path}.zip {output_path}")
-
-
-if __name__ == '__main__':
-    pack_firmware_image()
--- a/partitions.csv
+++ b/partitions.csv
@@ -4,6 +4,7 @@ nvs,      data, nvs,     0x9000,    0x4000,
 otadata,  data, ota,     0xd000,    0x2000,
 phy_init, data, phy,     0xf000,    0x1000,
 model,    data, spiffs,  0x10000,   0xF0000,
+storage,  data, spiffs,  0x100000,  1M,
 factory,  app,  factory, 0x200000,  4M,
 ota_0,    app,  ota_0,   0x600000,  4M,
 ota_1,    app,  ota_1,   0xA00000,  4M,
--- a/publish.py
+++ b/publish.py
@@ -1,51 +0,0 @@
-#! /usr/bin/env python3
-from dotenv import load_dotenv
-load_dotenv()
-
-import os
-import oss2
-import json
-
-def get_version():
-    with open('CMakeLists.txt', 'r') as f:
-        for line in f:
-            if line.startswith('set(PROJECT_VER'):
-                return line.split('"')[1]
-    return '0.0.0'
-
-def upload_bin_to_oss(bin_path, oss_key):
-    auth = oss2.Auth(os.environ['OSS_ACCESS_KEY_ID'], os.environ['OSS_ACCESS_KEY_SECRET'])
-    bucket = oss2.Bucket(auth, os.environ['OSS_ENDPOINT'], os.environ['OSS_BUCKET_NAME'])
-    bucket.put_object(oss_key, open(bin_path, 'rb'))
-
-
-if __name__ == '__main__':
-    # 获取版本号
-    version = get_version()
-    print(f'version: {version}')
-
-    # 上传 bin 文件到 OSS
-    upload_bin_to_oss('build/xiaozhi.bin', f'firmwares/xiaozhi-{version}.bin')
-
-    # File URL
-    file_url = os.path.join(os.environ['OSS_BUCKET_URL'], f'firmwares/xiaozhi-{version}.bin')
-    print(f'Uploaded bin to OSS: {file_url}')
-
-    firmware_json = {
-        "version": version,
-        "url": file_url
-    }
-    with open(f"build/firmware.json", "w") as f:
-        json.dump(firmware_json, f, indent=4)
-    
-    # copy firmware.json to server
-    firmware_config_path = os.environ['FIRMWARE_CONFIG_PATH']
-    ret = os.system(f'scp build/firmware.json {firmware_config_path}')
-    if ret != 0:
-        print(f'Failed to copy firmware.json to server')
-        exit(1)
-    print(f'Copied firmware.json to server: {firmware_config_path}')
-
-
-
-    
--- a/versions.py
+++ b/versions.py
@@ -0,0 +1,158 @@
+#! /usr/bin/env python3
+from dotenv import load_dotenv
+load_dotenv()
+
+import os
+import struct
+import zipfile
+import oss2
+import json
+
+def get_chip_id_string(chip_id):
+    return {
+        0x0000: "esp32",
+        0x0002: "esp32s2",
+        0x0005: "esp32c3",
+        0x0009: "esp32s3",
+        0x000C: "esp32c2",
+        0x000D: "esp32c6",
+        0x0010: "esp32h2",
+        0x0011: "esp32c5",
+        0x0012: "esp32p4",
+        0x0017: "esp32c5",
+    }[chip_id]
+
+def get_flash_size(flash_size):
+    MB = 1024 * 1024
+    return {
+        0x00: 1 * MB,
+        0x01: 2 * MB,
+        0x02: 4 * MB,
+        0x03: 8 * MB,
+        0x04: 16 * MB,
+        0x05: 32 * MB,
+        0x06: 64 * MB,
+        0x07: 128 * MB,
+    }[flash_size]
+
+def get_app_desc(data):
+    magic = struct.unpack("<I", data[0x00:0x04])[0]
+    if magic != 0xabcd5432:
+        raise Exception("Invalid app desc magic")
+    version = data[0x10:0x30].decode("utf-8").strip('\0')
+    project_name = data[0x30:0x50].decode("utf-8").strip('\0')
+    time = data[0x50:0x60].decode("utf-8").strip('\0')
+    date = data[0x60:0x70].decode("utf-8").strip('\0')
+    idf_ver = data[0x70:0x90].decode("utf-8").strip('\0')
+    elf_sha256 = data[0x90:0xb0].hex()
+    return {
+        "name": project_name,
+        "version": version,
+        "compile_time": date + "T" + time,
+        "idf_version": idf_ver,
+        "elf_sha256": elf_sha256,
+    }
+
+def get_board_name(folder):
+    basename = os.path.basename(folder)
+    if basename.startswith("v0.2"):
+        return "bread-simple"
+    if basename.startswith("v0.3") or basename.startswith("v0.4") or basename.startswith("v0.5") or basename.startswith("v0.6"):
+        if "ML307" in basename:
+            return "bread-compact-ml307"
+        else:
+            return "bread-compact-wifi"
+    raise Exception(f"Unknown board name: {basename}")
+
+def read_binary(dir_path):
+    merged_bin_path = os.path.join(dir_path, "merged-binary.bin")
+    data = open(merged_bin_path, "rb").read()[0x200000:]
+    if data[0] != 0xE9:
+        print(dir_path, "is not a valid image")
+        return
+    # get flash size
+    flash_size = get_flash_size(data[0x3] >> 4)
+    chip_id = get_chip_id_string(data[0xC])
+    # get segments
+    segment_count = data[0x1]
+    segments = []
+    offset = 0x18
+    for i in range(segment_count):
+        segment_size = struct.unpack("<I", data[offset + 4:offset + 8])[0]
+        offset += 8
+        segment_data = data[offset:offset + segment_size]
+        offset += segment_size
+        segments.append(segment_data)
+    assert offset < len(data), "offset is out of bounds"
+    
+    # extract bin file
+    bin_path = os.path.join(dir_path, "xiaozhi.bin")
+    if not os.path.exists(bin_path):
+        print("extract bin file to", bin_path)
+        open(bin_path, "wb").write(data)
+
+    # The app desc is in the first segment
+    desc = get_app_desc(segments[0])
+    return {
+        "chip_id": chip_id,
+        "flash_size": flash_size,
+        "board": get_board_name(dir_path),
+        "application": desc,
+    }
+
+def extract_zip(zip_path, extract_path):
+    if not os.path.exists(extract_path):
+        os.makedirs(extract_path)
+    print(f"Extracting {zip_path} to {extract_path}")
+    with zipfile.ZipFile(zip_path, 'r') as zip_ref:
+        zip_ref.extractall(extract_path)
+
+def upload_dir_to_oss(source_dir, target_dir):
+    auth = oss2.Auth(os.environ['OSS_ACCESS_KEY_ID'], os.environ['OSS_ACCESS_KEY_SECRET'])
+    bucket = oss2.Bucket(auth, os.environ['OSS_ENDPOINT'], os.environ['OSS_BUCKET_NAME'])
+    for filename in os.listdir(source_dir):
+        oss_key = os.path.join(target_dir, filename)
+        print('uploading', oss_key)
+        bucket.put_object(oss_key, open(os.path.join(source_dir, filename), 'rb'))
+
+def main():
+    release_dir = "releases"
+    versions = []
+    # look for zip files startswith "v"
+    for name in os.listdir(release_dir):
+        if name.startswith("v") and name.endswith(".zip"):
+            tag = name[:-4]
+            folder = os.path.join(release_dir, tag)
+            if not os.path.exists(folder):
+                os.makedirs(folder)
+                extract_zip(os.path.join(release_dir, name), folder)
+                info = read_binary(folder)
+                target_dir = os.path.join("firmwares", tag)
+                info["tag"] = tag
+                info["url"] = os.path.join(os.environ['OSS_BUCKET_URL'], target_dir, "xiaozhi.bin")
+                open(os.path.join(folder, "info.json"), "w").write(json.dumps(info, indent=4))
+                # upload all file to oss
+                upload_dir_to_oss(folder, target_dir)
+            # read info.json
+            info = json.load(open(os.path.join(folder, "info.json")))
+            versions.append(info)
+
+    # sort versions by version
+    versions.sort(key=lambda x: x["tag"], reverse=True)
+    # write versions to file
+    versions_path = os.path.join(release_dir, "versions.json")
+    open(versions_path, "w").write(json.dumps(versions, indent=4))
+    print(f"Versions written to {versions_path}")
+
+    # copy versions.json to server
+    versions_config_path = os.environ['VERSIONS_CONFIG_PATH']
+    ret = os.system(f'scp {versions_path} {versions_config_path}')
+    if ret != 0:
+        print(f'Failed to copy versions.json to server')
+        exit(1)
+    print(f'Copied versions.json to server: {versions_config_path}')
+
+
+
+if __name__ == "__main__":
+    main()
--- a/websocket.md
+++ b/websocket.md
@@ -1,160 +0,0 @@
-
-# AI 语音交互通信协议文档
-
-## 1. 连接建立与鉴权
-
-客户端通过 WebSocket 连接到服务器时，需要在 HTTP 头中包含以下信息：
-
- `Authorization`: Bearer token，格式为 "Bearer <access_token>"
- `Device-Id`: 设备 MAC 地址
- `Protocol-Version`: 协议版本号，当前为 2
-
-WebSocket URL: `wss://api.tenclass.net/xiaozhi/v1`
-
-## 2. 二进制数据
-
-客户端发送的二进制数据使用固定头格式的协议，如下：
-
-```cpp
-struct BinaryProtocol {
-    uint16_t version;        // 二进制协议版本，当前为 2
-    uint16_t type;           // 消息类型（0：音频流数据，1：JSON）
-    uint32_t reserved;       // 保留字段
-    uint32_t timestamp;      // 时间戳（保留用作回声消除，也可以用于UDP不可靠传输中的排序）
-    uint32_t payload_size;   // 负载大小
-    uint8_t payload[];       // 可以是音频数据（Opus 编码或协商的音频格式），也可以封装 JSON
-} __attribute__((packed));
-```
-
-注意：所有多字节整数字段使用网络字节序（大端序）。
-
-目前二进制数据跟 JSON 都是走同一个 WebSocket 连接，未来实时对话模式下，二进制音频数据可能走 UDP，可以扩展 hello 消息进行协商。
-
-## 3. 音频数据传输
-
- 客户端到服务器: 使用二进制协议发送 Opus 编码的音频数据
- 服务器到客户端: 使用二进制协议发送 Opus 编码的音频数据，格式与客户端发送的相同
-
-出现 payload_size 为 0 的音频数据包可以用做句子边界标记，可以忽略，但不要报错。
-
-## 4. 握手消息
-
-连接建立后，客户端发送一个 JSON 格式的 "hello" 消息，初始化服务器端的音频解码器。
-不需要等待服务器响应，随后即可发送音频数据。
-
-```json
-{
-  "type": "hello",
-  "response_mode": "auto",
-  "audio_params": {
-    "format": "opus",
-    "sample_rate": 16000,
-    "channels": 1
-  }
-}
-```
-
-应答模式 `response_mode` 可以为 `auto` 或 `manual`。
-
-`auto`：自动应答模式，服务器实时计算音频 VAD 并自动决定何时开始应答。
-
-`manual`：手动应答模式，客户端状态从 `listening` 变为 `idle` 时，服务器可以应答。
-
-## 5. 状态更新
-
-客户端在状态变化时发送 JSON 消息:
-
-```json
-{
-  "type": "state",
-  "state": "<新状态>"
-}
-```
-
-可能发送的状态值包括: `idle`, `wake_word_detected`, `listening`, `speaking`。
-
-示例:
-
-1、按住说话（`response_mode` 为 `manual`）
-
- 当按住说话按钮时，如果未连接服务器，则连接服务器，并编码、缓存当前音频数据，连接成功后，客户端设置状态为 `listening`，并在 hello 消息之后发送缓存的音频数据。
- 当按住说话按钮时，如果已连接服务器，则客户端设置状态为 `listening`，并发送音频数据。
- 当释放说话按钮时，状态变为 `idle`，此时服务器开始识别。
- 服务器开始应答时，推送 `stt` 和 `tts` 消息。
- 客户端开始播放音频时，状态设为 `speaking`。
- 客户端结束播放音频时，状态设为 `idle`。
- 在 `speaking` 状态下，按住说话按钮，会立即停止当前音频播放，状态变为 `listening`。
-
-2、语音唤醒，轮流对话（`response_mode` 为 `auto`）
-
- 连接服务器，发送 hello 消息，发送唤醒词音频数据，然后发送状态 `wake_word_detected`，服务器开始应答。
- 客户端开始播放音频时，状态设为 `speaking`，此时客户端不会发送音频数据。
- 客户端结束播放音频时，状态设为 `listening`，此时客户端发送音频数据。
- 服务器计算音频 VAD 自动选择时机开始应答时，推送 `stt` 和 `tts` 消息。
- 客户端收到 `tts`.`start` 时，开始播放音频，状态设为 `speaking`。
- 客户端收到 `tts`.`stop` 时，停止播放音频，状态设为 `listening`。
-
-3、语音唤醒，实时对话（`response_mode` 为 `real_time`）
-
- 连接服务器，发送 hello 消息，发送唤醒词音频数据，然后发送状态 `wake_word_detected`，服务器开始应答。
- 客户端开始播放音频时，状态设为 `speaking`。
- 客户端结束播放音频时，状态设为 `listening`。
- 在 `speaking` 和 `listening` 状态下，客户端都会发送音频数据。
- 服务器计算音频 VAD 自动选择时机开始应答时，推送 `stt` 和 `tts` 消息。
- 客户端收到 `stt` 时，状态设为 `listening`。如果当前有音频正在播放，则在当前 sentence 结束后停止播放音频。
- 客户端收到 `tts`.`start` 时，开始播放音频，状态设为 `speaking`。
- 客户端收到 `tts`.`stop` 时，停止播放音频，状态设为 `listening`。
-
-## 6. 服务器到客户端的消息
-
-### 6.1 语音识别结果 (STT)
-
-```json
-{
-  "type": "stt",
-  "text": "<识别出的文本>"
-}
-```
-
-### 6.2 文本转语音 (TTS)
-
-TTS开始:
-```json
-{
-  "type": "tts",
-  "state": "start",
-  "sample_rate": 24000
-}
-```
-
-句子开始:
-```json
-{
-  "type": "tts",
-  "state": "sentence_start",
-  "text": "你在干什么呀？"
-}
-```
-
-句子结束:
-```json
-{
-  "type": "tts",
-  "state": "sentence_end"
-}
-```
-
-TTS结束:
-```json
-{
-  "type": "tts",
-  "state": "stop"
-}
-```
-
-## 7. 连接管理
-
- 客户端检测到 WebSocket 断开连接时，应该停止音频播放并重置为空闲状态
- 在断开连接后，客户端按需重新发起连接（比如按钮按下或语音唤醒）
-
-这个文档概括了 WebSocket 通信协议的主要方面。
Author	SHA1	Message	Date
Terrence	91c7774117	bump to 0.6.2	2024-11-01 15:19:37 +08:00
Terrence	b0bc81b921	add new boards	2024-11-01 14:26:02 +08:00
Terrence	a701d5918e	add abort command	2024-10-31 05:57:13 +08:00
Terrence	6f5f5a0642	use protocol 3	2024-10-30 06:58:29 +08:00
Terrence	3e1e576272	update ml307 component version	2024-10-29 00:26:05 +08:00
Terrence	33518dca2b	add config files for known boards	2024-10-29 00:22:29 +08:00
Terrence	fe05a039a2	fix v0.4.x board name	2024-10-25 12:46:33 +08:00
Terrence	32a5977fcc	update to 0.4.1	2024-10-25 12:29:23 +08:00
Terrence	357cd95693	fix audio data overflow	2024-10-25 12:25:22 +08:00
Terrence	2833142870	update docs	2024-10-25 10:05:01 +08:00
Terrence	a2487f46c8	ES8311 + ES7210	2024-10-24 09:53:08 +08:00
Terrence	4c6da771ec	add versions.py for ota upgrade	2024-10-17 18:34:51 +08:00