DuoDuoProject/Doc/配置文档.md

### 一、服务器端下载
https://github.com/xinnan-tech/xiaozhi-esp32-server

### 二、安装$Anaconda3$

下载地址

> https://www.anaconda.com/download

将下面的四个目录添加到环境变量中

```
D:\anaconda3\Library\mingw-w64\bin
D:\anaconda3\Library\bin
D:\anaconda3\Library\usr\bin
D:\anaconda3\Scripts
```

确认是否安装成功

```
WIN+R，输入cmd打开控制台，输入conda -V
```



使用$Anaconda$ 的黑窗口进行配置

```
C:\ProgramData\Microsoft\Windows\Start Menu\Programs\Anaconda (anaconda3)\Anaconda Prompt.lnk
```

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/202504150744617.png)



设置$Anaconda$镜像

```cmd
conda config --remove-key channels
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/msys2/
conda config --show channels
```

**创建环境**

```
conda create --name xiaozhi-esp32-server python=3.10
```

**激活环境**

```
conda activate xiaozhi-esp32-server
```

### 三、部署步骤

**1、包安装**

```cmd
# 开始部署
d:
cd D:\dsWork\DuoDuoProject\xiaozhi-esp32-server-0.7.1\main\xiaozhi-server
pip install -r requirements.txt
```

**2、音频库安装**

```shell
# 音频的编码解码库
pip install opuslib
pip install ffmpeg
```

**安装音频处理工具，录制，转码，剪辑等**

https://github.com/ShiftMediaProject/opus/releases/download/v1.4/libopus_v1.4_msvc17.zip

依据下面的文档指示，对环境代码进行配置：

[解决Exception: Could not find Opus library. Make sure it is installed.](https://blog.csdn.net/Rysxt_/article/details/146361114)

``` python
# 找到报错文件 D:\dsWork\DuoDuoProject\xiaozhi-esp32-server-0.7.1\.venv\Lib\site-packages\opuslib_next\api\__init__.py

if lib_location is None:
    lib_location = r'D:\dsWork\DuoDuoProject\xiaozhi-esp32-server-0.7.1\opus.dll'
```

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721105355122.png)

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721115453134.png)

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721115414928.png)

**3、语音识别库安装**

下载语音识别模型文件

> https://pan.baidu.com/share/init?surl=QlgM58FHhYv1tFnUT_A8Sg&pwd=qvna

将model.pt 放到

```
D:\dsWork\DuoDuoProject\xiaozhi-esp32-server-0.7.1\main\xiaozhi-server\models\SenseVoiceSmall
```

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721104644515.png)

**4、配置文件**

**配置项目**

- 创建data目录
- 把源码目录下的config.yaml文件复制一份，重命名为.config.yaml

**5、Python端服务启动**

```cmd
# 启动后端服务【端口：8002】
D:\dsWork\DuoDuoProject\xiaozhi-esp32-server-0.7.1\main\xiaozhi-server\app.py
```

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721110521298.png)

**6、$JAVA$ $OTA$启动**

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721112548846.png)

**7、启动前端页面**【不启动也可以~】

```cmd
# 安装Node.js 

#我的电脑：
D:\dsWork\DuoDuoProject\xiaozhi-esp32-server-0.7.1\main\manager-web

#在地址栏输入 cmd 回车，然后输入：
npm install
npm run serve
```

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721144828729.png)

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721144851697.png)

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721144932309.png)

**8、测试页面**

```
file:///D:/dsWork/DuoDuoProject/xiaozhi-esp32-server-0.7.1/main/xiaozhi-server/test/test_page.html
```

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721112711634.png)

```cmd
http://127.0.0.1:8001/#/home
huanghai
DsideaL@123
```

### 四、Windows搭建 ESP IDF 5.3.2开发环境以及编译小智

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721131022851.png)

### 五、修改固件

- **固件编译工具**

`	esp-idf-tools-setup-offline-5.3.2.exe`



![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721132417926.png)



- **驱动程序**

​	`CH343SER.EXE`



![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721131517125.png)



- **实验材料**
  处理器型号  $ESP32-S3-N16R8$

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/202504150754335.png)

屏幕 $0.91$英寸，$SSD$ $1306$，分辨率：$128*32$

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/202504150753720.png)





- **烧录步骤**

​	**步骤1**
​	双击桌面上的ESP-IDF 5.3 PowerShell 快捷方式

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721133227523.png)

​	**步骤2**

​	https://icnynnzcwou8.feishu.cn/wiki/JEYDwTTALi5s2zkGlFGcDiRknXf

```sh
# 进入工作目录
cd D:\dsWork\DuoDuoProject\xiaozhi-esp32

# 全面清空
del build

# 构建一次
idf.py build
```

​	**步骤3**

​	**修改配置文件**

```
D:\dsWork\DuoDuoProject\xiaozhi-esp32\main\Kconfig.projbuild 
```

```yaml
# 设置OTA地址
config OTA_VERSION_URL
    string "OTA Version URL"
    default "http://172.16.1.114:8002/xiaozhi/ota/"
    help
        The application will access this URL to check for updates.

# 配置本地服务器的服务地址
config WEBSOCKET_URL
    depends on CONNECTION_TYPE_WEBSOCKET
    string "Websocket URL"
    default "ws://172.16.1.114:8000/xiaozhi/v1/"
    help
        Communication with the server through websocket after wake up.

# 修改默认连接方式
choice CONNECTION_TYPE
    prompt "Connection Type"
    default CONNECTION_TYPE_WEBSOCKET
    help
        网络数据传输协议
    config CONNECTION_TYPE_MQTT_UDP
        bool "MQTT + UDP"
    config CONNECTION_TYPE_WEBSOCKET
        bool "Websocket"
endchoice

# 设置默认的开发板类型
choice BOARD_TYPE
    prompt "Board Type"
    default BOARD_TYPE_BREAD_COMPACT_WIFI
    help
        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_WIFI_LCD
        bool "面包板新版接线（WiFi）+ LCD"
    config BOARD_TYPE_BREAD_COMPACT_ML307
        bool "面包板新版接线（ML307 AT）"
    config BOARD_TYPE_BREAD_COMPACT_ESP32
        bool "面包板（WiFi） ESP32 DevKit"
    config BOARD_TYPE_BREAD_COMPACT_ESP32_LCD
        bool "面包板（WiFi+ LCD） ESP32 DevKit"

# 设置默认屏幕
choice DISPLAY_OLED_TYPE
    depends on BOARD_TYPE_BREAD_COMPACT_WIFI || BOARD_TYPE_BREAD_COMPACT_ML307 || BOARD_TYPE_BREAD_COMPACT_ESP32
    prompt "OLED Type"
    default OLED_SSD1306_128X32
    help
        OLED 屏幕类型选择
    config OLED_SSD1306_128X32
        bool "SSD1306, 分辨率128*32"
    config OLED_SSD1306_128X64
        bool "SSD1306, 分辨率128*64"
    config OLED_SH1106_128X64
        bool "SH1106, 分辨率128*64"
endchoice
```

**设定目标板【默认就是，不用修改】**

```
d:
cd D:\dsWork\QingLong\XiaoZhi\xiaozhi-esp32
idf.py set-target esp32s3
```

**进入配置界面**

```
idf.py menuconfig
```

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/202504150817323.png)配置一下

进入$Xiaozhi$ $Assistant$选项

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/202504150817839.png)

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/202504150818535.png)

$Board$ $Type$: 

    config BOARD_TYPE_BREAD_COMPACT_WIFI
        bool "面包板新版接线（WiFi）"       

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/202504150812722.png)

$DISPLAY\_OLED\_TYPE$:

$OLED\_SSD1306\_128X32$

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/202504150757608.png)

按ESC退回主界面，再点S进行保存。

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/202504150822282.png)

**编译固件**

```
idf.py build
```

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/202504150834785.png)

**打包$bin$固件**

```
cd scripts
python release.py
```

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/202504150835200.png)

**打包后的产物**

```
D:\dsWork\DuoDuoProject\xiaozhi-esp32\build\merged-binary.bin
```

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721142527263.png)



- 开始烧录

  ![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721142945436.png)

  ![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721143146062.png)![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250721143213622.png)

  烧录完成后，按下主板上的RST按钮，也就是板子右侧上方第一个按钮。

  然后进入配网模式：
  查看显示屏上的热点名称，即SSID值，用手机或电脑连接到此热点，对ESP32主板进行配网，让其连入WIFI

### 六、网络映射

网络映射

知识库
8000  --->8100

小智
8000 WEB UI   	--> 27011
8002 OTA+JAVA      --> 27012
8003 http server     --> 27013
8004 websocket     --> 27014



需要注意的是绘制科技的服务器，需要配置一下安全策略的范围：

![](https://dsideal.obs.cn-north-1.myhuaweicloud.com/HuangHai/BlogImages/%7Byear%7D/%7Bmonth%7D/%7Bmd5%7D.%7BextName%7D/20250722075509715.png)

测试页面
file:///D:/dsWork/QingLong/XiaoZhi/xiaozhi-esp32-server/main/xiaozhi-server/test/test_page.html

### 七、相关资料

1. [固件开源地址](https://github.com/78/xiaozhi-esp32)

2. [服务端开源地址](https://github.com/xinnan-tech/xiaozhi-esp32-server)

3. [Windows搭建 ESP IDF 5.4.1开发环境以及编译小智](https://icnynnzcwou8.feishu.cn/wiki/JEYDwTTALi5s2zkGlFGcDiRknXf)


4. [小智 AI 聊天机器人百科全书](
   https://ccnphfhqs21z.feishu.cn/wiki/F5krwD16viZoF0kKkvDcrZNYnhb)
5. [购买店铺链接](https://docs.qq.com/sheet/DWmRhbnVHSnRWc2xz?tab=vajqu5)
6. [开源小智Server视频教程](https://space.bilibili.com/2042143558/lists/4859662?type=season)
7. [编译固件方法](https://github.com/xinnan-tech/xiaozhi-esp32-server/blob/main/docs/firmware-build.md)



D:\dsWork\DuoDuoProject\xiaozhi-esp32-server-0.7.1\main\xiaozhi-server\core\connection.py

681行：

```python
 # 黄海扩展
            if '史记' in query:
                llm_responses = self.llm.response(
                    self.session_id,
                    self.dialogue.get_llm_dialogue_with_memory(
                        query, self.config.get("voiceprint", {})
                    ),
                )
```