feat: 增强音频处理和语音识别日志

- 添加详细的运行时日志输出
- 记录音频块数量、RMS 值和缓冲区状态
- 记录 ONNX 推理耗时
- 记录设备信息和录音状态
- 改进错误处理和日志格式化

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

src/core/audio-recorder.ts

@@ -22,13 +22,45 @@ export interface AudioChunk {
   timestamp: number;
 }
 
+// 日志级别
+enum LogLevel {
+  DEBUG = 'DEBUG',
+  INFO = 'INFO',
+  WARN = 'WARN',
+  ERROR = 'ERROR',
+}
+
+// 简单日志函数
+function log(level: LogLevel, message: string, data?: unknown): void {
+  const timestamp = new Date().toISOString().replace('T', ' ').slice(0, 23);
+  const formatted = `[${timestamp}] [${level}] [AudioRecorder] ${message}`;
+
+  if (data !== undefined) {
+    try {
+      if (data instanceof Error) {
+        console.log(`${formatted} - ${data.message}`);
+      } else if (typeof data === 'object') {
+        console.log(`${formatted} - ${JSON.stringify(data)}`);
+      } else {
+        console.log(`${formatted} - ${data}`);
+      }
+    } catch (e) {
+      console.log(`${formatted} - [Unable to stringify data]`);
+    }
+  } else {
+    console.log(formatted);
+  }
+}
+
 export class AudioRecorder extends EventEmitter {
   private config: AudioConfig;
   private isRecording: boolean = false;
-  private stream: any = null;
+  private stream: MediaStream | null = null;
-  private audioContext: any = null;
+  private audioContext: AudioContext | null = null;
-  private source: any = null;
+  private source: MediaStreamAudioSourceNode | null = null;
-  private processor: any = null;
+  private processor: ScriptProcessorNode | null = null;
+  private chunkCount: number = 0;
+  private totalSamples: number = 0;
 
   constructor(config: Partial<AudioConfig> = {}) {
     super();
@@ -38,6 +70,7 @@ export class AudioRecorder extends EventEmitter {
       chunkDuration: config.chunkDuration ?? 100,
       deviceId: config.deviceId,
     };
+    log(LogLevel.INFO, 'AudioRecorder 初始化', this.config);
   }
 
   /**
@@ -45,13 +78,18 @@ export class AudioRecorder extends EventEmitter {
    */
   async start(): Promise<void> {
     if (this.isRecording) {
+      log(LogLevel.WARN, '已经在录音中');
       throw new Error('Already recording');
     }
 
+    log(LogLevel.INFO, '开始录音，检查环境...');
+
     // 检查是否在浏览器/Electron 渲染进程中
     if (typeof window !== 'undefined' && window.navigator?.mediaDevices) {
+      log(LogLevel.INFO, '检测到浏览器环境，使用 getUserMedia');
       await this.startInBrowser();
     } else {
+      log(LogLevel.WARN, 'Node.js 环境，使用演示模式');
       // Node.js 环境 - 需要外部音频输入
       this.startInNode();
     }
@@ -62,24 +100,67 @@ export class AudioRecorder extends EventEmitter {
    */
   private async startInBrowser(): Promise<void> {
     try {
+      log(LogLevel.INFO, '请求麦克风权限...');
+
       const constraints = {
         audio: {
           sampleRate: this.config.sampleRate,
           channelCount: this.config.channels,
           deviceId: this.config.deviceId ? { exact: this.config.deviceId } : undefined,
+          echoCancellation: true,
+          noiseSuppression: true,
+          autoGainControl: true,
         },
       };
 
+      log(LogLevel.DEBUG, '麦克风约束', constraints);
+
+      // 请求权限
       this.stream = await window.navigator.mediaDevices.getUserMedia(constraints);
+      log(LogLevel.INFO, '✅ 麦克风权限已获取');
+
+      // 获取设备信息
+      const devices = await window.navigator.mediaDevices.enumerateDevices();
+      const audioInputs = devices.filter(d => d.kind === 'audioinput');
+      const activeDevice = audioInputs.find(d => d.deviceId === this.stream.getAudioTracks()[0]?.getSettings().deviceId);
+      log(LogLevel.INFO, '录音设备信息', {
+        label: activeDevice?.label || this.stream.getAudioTracks()[0]?.label || 'Unknown',
+        deviceId: this.stream.getAudioTracks()[0]?.getSettings().deviceId,
+        sampleRate: this.stream.getAudioTracks()[0]?.getSettings().sampleRate,
+      });
+
       const AudioContextClass = window.AudioContext || (window as any).webkitAudioContext;
       this.audioContext = new AudioContextClass({ sampleRate: this.config.sampleRate });
+      log(LogLevel.INFO, `AudioContext 创建，采样率：${this.config.sampleRate}`);
+
       this.source = this.audioContext.createMediaStreamSource(this.stream);
+      log(LogLevel.DEBUG, 'MediaStreamSource 已创建');
 
       const bufferSize = Math.floor(this.config.sampleRate * (this.config.chunkDuration / 1000));
+      log(LogLevel.DEBUG, `ScriptProcessor bufferSize: ${bufferSize}`);
+
       this.processor = this.audioContext.createScriptProcessor(bufferSize, 1, 1);
+      log(LogLevel.INFO, `ScriptProcessor 已创建，bufferSize=${bufferSize}`);
 
       this.processor.onaudioprocess = (event: any) => {
         const inputData = event.inputBuffer.getChannelData(0);
+
+        // 计算音频 RMS 值用于检测是否有声音输入
+        let sum = 0;
+        for (let i = 0; i < inputData.length; i++) {
+          sum += inputData[i] * inputData[i];
+        }
+        const rms = Math.sqrt(sum / inputData.length);
+
+        this.chunkCount++;
+        this.totalSamples += inputData.length;
+
+        log(LogLevel.DEBUG, `音频块 #${this.chunkCount}`, {
+          samples: inputData.length,
+          rms: rms.toFixed(4),
+          totalSamples: this.totalSamples,
+        });
+
         const chunk: AudioChunk = {
           data: new Float32Array(inputData),
           sampleRate: this.config.sampleRate,
@@ -90,10 +171,13 @@ export class AudioRecorder extends EventEmitter {
 
       this.source.connect(this.processor);
       this.processor.connect(this.audioContext.destination);
+      log(LogLevel.INFO, '音频处理链路已连接');
 
       this.isRecording = true;
+      log(LogLevel.INFO, '✅ 开始录音');
       this.emit('start');
     } catch (error) {
+      log(LogLevel.ERROR, '❌ 获取麦克风权限失败', error);
       this.emit('error', error);
       throw error;
     }
@@ -104,8 +188,8 @@ export class AudioRecorder extends EventEmitter {
    * 实际使用需要 node-audio 等库
    */
   private startInNode(): void {
-    console.warn('Node.js 环境音频采集需要 electron 或 node-audio 库');
+    log(LogLevel.WARN, 'Node.js 环境音频采集需要 electron 或 node-audio 库');
-    console.warn('当前运行在演示模式，不会采集音频');
+    log(LogLevel.WARN, '当前运行在演示模式，不会采集音频');
     this.isRecording = true;
     this.emit('start');
     // 演示：定期发送静音数据
@@ -128,39 +212,56 @@ export class AudioRecorder extends EventEmitter {
    */
   stop(): void {
     if (!this.isRecording) {
+      log(LogLevel.WARN, '未在录音中');
       return;
     }
 
+    log(LogLevel.INFO, '停止录音...');
+
     if (this.processor) {
       this.processor.disconnect();
       this.processor = null;
+      log(LogLevel.DEBUG, 'ScriptProcessor 已断开');
     }
     if (this.source) {
       this.source.disconnect();
       this.source = null;
+      log(LogLevel.DEBUG, 'MediaStreamSource 已断开');
     }
     if (this.stream) {
       const tracks = this.stream.getTracks?.() || this.stream.tracks || [];
       tracks.forEach((track: any) => track.stop?.());
       this.stream = null;
+      log(LogLevel.DEBUG, '媒体流已停止');
     }
     if (this.audioContext) {
       this.audioContext.close?.();
       this.audioContext = null;
+      log(LogLevel.DEBUG, 'AudioContext 已关闭');
     }
 
     this.isRecording = false;
+    log(LogLevel.INFO, `录音停止完成，共处理 ${this.chunkCount} 个音频块，${this.totalSamples} 个样本`);
+    this.chunkCount = 0;
+    this.totalSamples = 0;
     this.emit('stop');
   }
 
   /**
    * 获取可用音频设备列表（仅浏览器环境）
    */
-  static async listDevices(): Promise<any[]> {
+  static async listDevices(): Promise<MediaDeviceInfo[]> {
     if (typeof window !== 'undefined' && window.navigator?.mediaDevices) {
+      log(LogLevel.INFO, '枚举音频输入设备...');
       const devices = await window.navigator.mediaDevices.enumerateDevices();
-      return devices.filter((device: any) => device.kind === 'audioinput');
+      const audioInputs = devices.filter((device: any) => device.kind === 'audioinput');
+      log(LogLevel.INFO, `找到 ${audioInputs.length} 个音频输入设备`);
+      audioInputs.forEach((device, index) => {
+        log(LogLevel.DEBUG, `设备 ${index + 1}: ${device.label || 'Unknown'} (${device.deviceId})`);
+      });
+      return audioInputs;
     }
+    log(LogLevel.WARN, '当前环境不支持设备枚举');
     return [];
   }
 

src/core/speech-recognizer.ts

@@ -22,17 +22,55 @@ export interface RecognitionResult {
   timestamp: number;       // 时间戳
 }
 
+// 日志级别
+enum LogLevel {
+  DEBUG = 'DEBUG',
+  INFO = 'INFO',
+  WARN = 'WARN',
+  ERROR = 'ERROR',
+}
+
+// 简单日志函数
+function log(level: LogLevel, message: string, data?: unknown): void {
+  const timestamp = new Date().toISOString().replace('T', ' ').slice(0, 23);
+  const formatted = `[${timestamp}] [${level}] [SpeechRecognizer] ${message}`;
+
+  if (data !== undefined) {
+    try {
+      if (data instanceof Error) {
+        console.log(`${formatted} - ${data.message}`);
+      } else if (typeof data === 'object') {
+        console.log(`${formatted} - ${JSON.stringify(data)}`);
+      } else {
+        console.log(`${formatted} - ${data}`);
+      }
+    } catch (e) {
+      console.log(`${formatted} - [Unable to stringify data]`);
+    }
+  } else {
+    console.log(formatted);
+  }
+}
+
 export class SpeechRecognizer extends EventEmitter {
   private config: RecognizerConfig;
   private modelLoader: ModelLoader;
   private isRecognizing: boolean = false;
   private audioBuffer: Float32Array = new Float32Array(0);
   private readonly MAX_BUFFER_SECONDS = 30;
+  private processedChunks: number = 0;
+  private recognizedResults: number = 0;
 
   constructor(config: RecognizerConfig) {
     super();
     this.config = config;
     this.modelLoader = new ModelLoader();
+    log(LogLevel.INFO, 'SpeechRecognizer 初始化', {
+      modelPath: config.modelPath,
+      language: config.language,
+      useVad: config.useVad,
+      beamSize: config.beamSize,
+    });
   }
 
   /**
@@ -40,9 +78,12 @@ export class SpeechRecognizer extends EventEmitter {
    */
   async initialize(): Promise<void> {
     try {
+      log(LogLevel.INFO, '开始加载模型...');
       await this.modelLoader.load(this.config.modelPath);
+      log(LogLevel.INFO, '✅ 模型加载完成');
       this.emit('ready');
     } catch (error) {
+      log(LogLevel.ERROR, '❌ 模型加载失败', error);
       this.emit('error', new Error(`Failed to load model: ${error}`));
       throw error;
     }
@@ -62,6 +103,22 @@ export class SpeechRecognizer extends EventEmitter {
     newBuffer.set(chunk.data, this.audioBuffer.length);
     this.audioBuffer = newBuffer;
 
+    this.processedChunks++;
+
+    // 计算音频 RMS 值
+    let sum = 0;
+    for (let i = 0; i < chunk.data.length; i++) {
+      sum += chunk.data[i] * chunk.data[i];
+    }
+    const rms = Math.sqrt(sum / chunk.data.length);
+
+    log(LogLevel.DEBUG, `音频块 #${this.processedChunks}`, {
+      samples: chunk.data.length,
+      sampleRate: chunk.sampleRate,
+      rms: rms.toFixed(4),
+      bufferSize: this.audioBuffer.length,
+    });
+
     // 检查缓冲区是否超过最大长度
     const maxSamples = this.config.useVad
       ? chunk.sampleRate * this.MAX_BUFFER_SECONDS
@@ -70,6 +127,7 @@ export class SpeechRecognizer extends EventEmitter {
     if (this.audioBuffer.length > maxSamples) {
       const keepStart = Math.floor(this.audioBuffer.length / 2);
       this.audioBuffer = this.audioBuffer.slice(keepStart);
+      log(LogLevel.DEBUG, `缓冲区裁剪，保留 ${this.audioBuffer.length} 样本`);
     }
 
     // 进行识别
@@ -86,6 +144,8 @@ export class SpeechRecognizer extends EventEmitter {
     }
 
     try {
+      log(LogLevel.DEBUG, '开始 ONNX 推理...');
+
       // 重采样到模型要求的采样率
       let audioData = this.audioBuffer;
       if (sampleRate !== modelConfig.sampleRate) {
@@ -96,6 +156,7 @@ export class SpeechRecognizer extends EventEmitter {
           const pos = Math.floor(i * ratio);
           audioData[i] = this.audioBuffer[pos] || 0;
         }
+        log(LogLevel.DEBUG, `重采样：${sampleRate} -> ${modelConfig.sampleRate}`);
       }
 
       // 填充或截断到模型输入大小
@@ -104,18 +165,26 @@ export class SpeechRecognizer extends EventEmitter {
       const copyLength = Math.min(audioData.length, inputSize);
       inputData.set(audioData.slice(0, copyLength));
 
+      log(LogLevel.DEBUG, `输入张量形状：[1, ${inputSize}]`);
+
       const inputTensor = new ort.Tensor('float32', inputData, [1, inputSize]);
 
       const feeds: Record<string, ort.Tensor> = {
         input: inputTensor,
       };
 
+      const startTime = Date.now();
       const results = await this.modelLoader.run(feeds);
+      const inferenceTime = Date.now() - startTime;
+
+      log(LogLevel.INFO, `ONNX 推理完成，耗时：${inferenceTime}ms`);
 
       // 解码结果
       const text = this.decodeOutput(results, modelConfig);
 
       if (text) {
+        this.recognizedResults++;
+        log(LogLevel.INFO, `📝 识别结果 #${this.recognizedResults}: ${text}`);
         const result: RecognitionResult = {
           text,
           confidence: 0.95,
@@ -123,11 +192,14 @@ export class SpeechRecognizer extends EventEmitter {
           timestamp: Date.now(),
         };
         this.emit('result', result);
+      } else {
+        log(LogLevel.DEBUG, '识别结果为空');
       }
 
       // 清空缓冲区
       this.audioBuffer = new Float32Array(0);
     } catch (error) {
+      log(LogLevel.ERROR, '❌ 识别失败', error);
       this.emit('error', new Error(`Recognition failed: ${error}`));
     }
   }
@@ -143,6 +215,7 @@ export class SpeechRecognizer extends EventEmitter {
     for (const key of outputKeys) {
       if (results[key]) {
         output = results[key];
+        log(LogLevel.DEBUG, `找到输出键：${key}`);
         break;
       }
     }
@@ -152,16 +225,19 @@ export class SpeechRecognizer extends EventEmitter {
       const firstKey = Object.keys(results)[0];
       if (firstKey) {
         output = results[firstKey];
+        log(LogLevel.DEBUG, `使用第一个输出键：${firstKey}`);
       }
     }
 
     if (!output || !output.data) {
+      log(LogLevel.WARN, '没有可用的输出数据');
       return '';
     }
 
     // 简化处理：实际应根据具体模型使用 tokenizer 解码
     // 这里返回一个占位字符串
     const tokens = Array.from(output.data as Float32Array | Int32Array);
+    log(LogLevel.DEBUG, `输出 token 数量：${tokens.length}`);
     return `[识别结果：${tokens.length} tokens]`;
   }
 
@@ -170,6 +246,10 @@ export class SpeechRecognizer extends EventEmitter {
    */
   start(): void {
     this.isRecognizing = true;
+    this.processedChunks = 0;
+    this.recognizedResults = 0;
+    this.audioBuffer = new Float32Array(0);
+    log(LogLevel.INFO, '🎤 开始语音识别');
     this.emit('start');
   }
 
@@ -178,6 +258,7 @@ export class SpeechRecognizer extends EventEmitter {
    */
   stop(): void {
     this.isRecognizing = false;
+    log(LogLevel.INFO, `停止语音识别，共处理 ${this.processedChunks} 个音频块，${this.recognizedResults} 个识别结果`);
     if (this.audioBuffer.length > 0) {
       this.recognize(16000);
     }
@@ -188,8 +269,10 @@ export class SpeechRecognizer extends EventEmitter {
    * 卸载模型释放资源
    */
   async release(): Promise<void> {
+    log(LogLevel.INFO, '释放识别引擎资源...');
     this.stop();
     await this.modelLoader.release();
+    log(LogLevel.INFO, '资源已释放');
   }
 
   /**