语音控制集成

语音控制集成

概述

本节介绍如何为智能家居面板添加语音控制功能。学习完成后，您将能够：

• 理解 ESP32 语音控制的技术方案
• 实现简单的语音命令识别和响应
• 将语音命令转换为 MQTT 控制指令
• 评估语音控制方案的技术边界

前置要求

在开始本节之前，请确保：

• 已完成 MQTT 通信配置
• 了解基础的语音识别概念
• I2S 麦克风模块已接线

Voice Control Options

Solution Comparison

方案	方式	优点	缺点	成本
离线语音识别	ESP32-SR 模块	无需网络、延迟低	命令有限	$5-8
在线语音识别	云端 API（百度/阿里）	识别准确率高	需要网络、有延迟	API 费用
语音唤醒 + 命令	离线唤醒 + 在线识别	低功耗、高准确率	实现复杂	$8-15
外部语音助手	对接 Alexa/Google Home	生态成熟	需要额外设备	$30-50

ESP32-SR Voice Recognition

Hardware Integration

ESP32-SR 是专用的语音识别模块，通过 UART 与 ESP32 通信：

#include <SoftwareSerial.h>

// ESP32-SR 连接
#define SR_RX 16
#define SR_TX 17
#define SR_BAUD 9600

SoftwareSerial srSerial(SR_RX, SR_TX);

class VoiceRecognition {
private:
  // 命令 ID 映射
  struct CommandEntry {
    int id;
    const char* name;
    const char* mqttTopic;
    const char* mqttPayload;
  };

  CommandEntry commands[] = {
    {1, "打开灯光", "home/light/livingroom/set", "{\"state\":\"ON\"}"},
    {2, "关闭灯光", "home/light/livingroom/set", "{\"state\":\"OFF\"}"},
    {3, "打开空调", "home/climate/livingroom/mode/set", "\"cool\""},
    {4, "关闭空调", "home/climate/livingroom/mode/set", "\"off\""},
    {5, "离家模式", "home/scene/livingroom/trigger", "\"离家\""},
    {6, "回家模式", "home/scene/livingroom/trigger", "\"回家\""},
    {7, "睡眠模式", "home/scene/livingroom/trigger", "\"睡眠\""},
    {8, "温度设为26度", "home/climate/livingroom/temperature_set", "\"26\""},
  };

public:
  bool begin(int rx, int tx) {
    srSerial.begin(SR_BAUD, SWSERIAL_8N1, rx, tx);
    delay(1000);

    // 配置命令列表
    for (const auto& cmd : commands) {
      srSerial.print("CMD_SET,");
      srSerial.print(cmd.id);
      srSerial.print(",");
      srSerial.println(cmd.name);
      delay(100);
    }

    srSerial.println("SAVE");  // 保存配置
    return true;
  }

  // 检查语音识别结果
  int checkCommand() {
    if (srSerial.available()) {
      String response = srSerial.readStringUntil('\n');
      response.trim();

      if (response.startsWith("CMD_RECV,")) {
        int cmdId = response.substring(9).toInt();
        return cmdId;
      }
    }
    return -1;
  }

  // 执行语音命令
  void executeCommand(int cmdId, PubSubClient* mqtt) {
    for (const auto& cmd : commands) {
      if (cmd.id == cmdId) {
        Serial.printf("语音命令: %s\n", cmd.name);
        mqtt->publish(cmd.mqttTopic, cmd.mqttPayload);

        // 刷新 UI 显示
        update_voice_feedback(cmd.name);
        return;
      }
    }
  }
};

VoiceRecognition voice;

MQTT Voice Integration

Voice Command via MQTT

另一种方案：外部语音助手（如 Amazon Alexa、Google Home）通过 MQTT 将语音指令转发到系统：

// Node-RED Flow: 语音命令到 MQTT
[
  {
    "id": "voice_command_in",
    "type": "mqtt in",
    "topic": "home/voice/command",
    "server": "localhost"
  },
  {
    "id": "voice_parser",
    "type": "function",
    "func": `
      var text = msg.payload.text || msg.payload;

      // 关键词匹配
      var commands = [
        {pattern: /打开.*灯/, topic: "home/light/livingroom/set", payload: '{"state":"ON"}'},
        {pattern: /关闭.*灯/, topic: "home/light/livingroom/set", payload: '{"state":"OFF"}'},
        {pattern: /空调.*26/, topic: "home/climate/livingroom/temperature_set", payload: '26'},
        {pattern: /离家/, topic: "home/scene/livingroom/trigger", payload: '"离家"'},
        {pattern: /回家/, topic: "home/scene/livingroom/trigger", payload: '"回家"'},
        {pattern: /温度.*高点?/, topic: "home/climate/livingroom/temperature_set", payload: '26'},
        {pattern: /温度.*低点?/, topic: "home/climate/livingroom/temperature_set", payload: '22'}
      ];

      for (var cmd of commands) {
        if (cmd.pattern.test(text)) {
          return {topic: cmd.topic, payload: cmd.payload};
        }
      }

      // 未识别
      return {topic: "home/voice/response", payload: "抱歉，无法识别该命令"};
    `
  },
  {
    "id": "voice_command_out",
    "type": "mqtt out",
    "qos": 1,
    "server": "localhost"
  }
]

LVGL Voice Feedback

Voice Recognition UI

// 语音反馈 UI
lv_obj_t* voiceIndicator;
lv_obj_t* voiceFeedback;
lv_timer_t* voiceTimer;

void create_voice_ui() {
  // 语音状态指示器
  voiceIndicator = lv_label_create(screen);
  lv_label_set_text(voiceIndicator, "🎤");
  lv_obj_set_style_text_font(voiceIndicator, &lv_font_montserrat_24, 0);
  lv_obj_align(voiceIndicator, LV_ALIGN_BOTTOM_RIGHT, -20, -20);

  // 语音反馈标签
  voiceFeedback = lv_label_create(screen);
  lv_label_set_text(voiceFeedback, "");
  lv_obj_set_style_text_font(voiceFeedback, &lv_font_montserrat_16, 0);
  lv_obj_set_style_text_color(voiceFeedback, lv_palette_main(LV_PALETTE_BLUE), 0);
  lv_obj_align(voiceFeedback, LV_ALIGN_BOTTOM_MID, 0, -50);
}

void update_voice_feedback(const char* command) {
  lv_label_set_text_fmt(voiceFeedback, "语音: %s", command);
  lv_obj_clear_flag(voiceFeedback, LV_OBJ_FLAG_HIDDEN);

  // 3 秒后自动消失
  if (voiceTimer) lv_timer_del(voiceTimer);
  voiceTimer = lv_timer_create([](lv_timer_t* t) {
    lv_obj_add_flag(voiceFeedback, LV_OBJ_FLAG_HIDDEN);
  }, 3000, NULL);
  lv_timer_set_repeat_count(voiceTimer, 1);
}

// PIR + 语音反馈：有人靠近时显示语音提示
void check_proximity_voice() {
  if (sensors.motionDetected && !isVoicePromptShown) {
    lv_label_set_text(voiceFeedback, "说"你好"唤醒语音控制");
    lv_obj_clear_flag(voiceFeedback, LV_OBJ_FLAG_HIDDEN);
    isVoicePromptShown = true;
  }

  // 10 秒后隐藏提示
  if (isVoicePromptShown && millis() - lastMotionTime > 10000) {
    lv_obj_add_flag(voiceFeedback, LV_OBJ_FLAG_HIDDEN);
    isVoicePromptShown = false;
  }
}

Technical Limitations

Voice Control Constraints

限制因素	离线方案	在线方案
命令数量	20-50 条	无限制
识别准确率	80-90%	95-98%
响应延迟	0.5-1s	1-3s（含网络）
抗噪能力	一般（<60dB）	良好（使用降噪）
多语言	需定制	原生支持
离线运行	✅ 支持	❌ 需要网络

Pre-sales Key Points

语音控制价值

能力	说明	买家沟通要点
免提控制	无需触摸面板	”做饭时你说’关灯’，面板自动响应”
场景快速	一句话触发场景	”说’睡觉了’，所有灯关闭窗帘落下”
无障碍	老人儿童友好	”不用学习App，说话就能控制”
离线可用	不依赖云服务	”即使断网，语音控制依然可用”

Summary

本节介绍了智能家居面板的语音控制集成：

1. 方案选择：离线语音识别（ESP32-SR）vs 在线语音 API
2. 关键词匹配：Node-RED 进行语音命令的关键词解析
3. LVGL 反馈：语音控制状态和命令反馈的 UI 显示
4. MQTT 集成：语音命令通过 MQTT 转换为设备控制指令
5. 技术边界：离线方案命令有限但延迟低，在线方案准确率高但依赖网络