@lokutor/sdk 1.1.2 → 1.1.8

package/dist/index.js

@@ -21,11 +21,22 @@ var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: tru
 var index_exports = {};
 __export(index_exports, {
   AUDIO_CONFIG: () => AUDIO_CONFIG,
+  BrowserAudioManager: () => BrowserAudioManager,
   DEFAULT_URLS: () => DEFAULT_URLS,
   Language: () => Language,
+  StreamResampler: () => StreamResampler,
   TTSClient: () => TTSClient,
   VoiceAgentClient: () => VoiceAgentClient,
   VoiceStyle: () => VoiceStyle,
+  applyLowPassFilter: () => applyLowPassFilter,
+  bytesToPcm16: () => bytesToPcm16,
+  calculateRMS: () => calculateRMS,
+  float32ToPcm16: () => float32ToPcm16,
+  normalizeAudio: () => normalizeAudio,
+  pcm16ToBytes: () => pcm16ToBytes,
+  pcm16ToFloat32: () => pcm16ToFloat32,
+  resample: () => resample,
+  resampleWithAntiAliasing: () => resampleWithAntiAliasing,
   simpleConversation: () => simpleConversation,
   simpleTTS: () => simpleTTS
 });
@@ -86,10 +97,13 @@ var VoiceAgentClient = class {
   onTranscription;
   onResponse;
   onAudioCallback;
+  onVisemesCallback;
   onStatus;
   onError;
   isConnected = false;
   messages = [];
+  visemeListeners = [];
+  wantVisemes = false;
   constructor(config) {
     this.apiKey = config.apiKey;
     this.prompt = config.prompt;
@@ -98,8 +112,10 @@ var VoiceAgentClient = class {
     this.onTranscription = config.onTranscription;
     this.onResponse = config.onResponse;
     this.onAudioCallback = config.onAudio;
+    this.onVisemesCallback = config.onVisemes;
     this.onStatus = config.onStatus;
     this.onError = config.onError;
+    this.wantVisemes = config.visemes || false;
   }
   /**
    * Connect to the Lokutor Voice Agent server
@@ -151,7 +167,8 @@ var VoiceAgentClient = class {
     this.ws.send(JSON.stringify({ type: "prompt", data: this.prompt }));
     this.ws.send(JSON.stringify({ type: "voice", data: this.voice }));
     this.ws.send(JSON.stringify({ type: "language", data: this.language }));
-    console.log(`\u2699\uFE0F Configured: voice=${this.voice}, language=${this.language}`);
+    this.ws.send(JSON.stringify({ type: "visemes", data: this.wantVisemes }));
+    console.log(`\u2699\uFE0F Configured: voice=${this.voice}, language=${this.language}, visemes=${this.wantVisemes}`);
   }
   /**
    * Send raw PCM audio data to the server
@@ -206,6 +223,11 @@ var VoiceAgentClient = class {
           };
           console.log(`${icons[msg.data] || ""} Status: ${msg.data}`);
           break;
+        case "visemes":
+          if (Array.isArray(msg.data) && msg.data.length > 0) {
+            this.emit("visemes", msg.data);
+          }
+          break;
         case "error":
           if (this.onError) this.onError(msg.data);
           console.error(`\u274C Server error: ${msg.data}`);
@@ -219,11 +241,17 @@ var VoiceAgentClient = class {
     if (event === "audio") {
       if (this.onAudioCallback) this.onAudioCallback(data);
       this.audioListeners.forEach((l) => l(data));
+    } else if (event === "visemes") {
+      if (this.onVisemesCallback) this.onVisemesCallback(data);
+      this.visemeListeners.forEach((l) => l(data));
     }
   }
   onAudio(callback) {
     this.audioListeners.push(callback);
   }
+  onVisemes(callback) {
+    this.visemeListeners.push(callback);
+  }
   /**
    * Disconnect from the server
    */
@@ -330,14 +358,435 @@ async function simpleTTS(options) {
   const client = new TTSClient({ apiKey: options.apiKey });
   return client.synthesize(options);
 }
+
+// src/audio-utils.ts
+function pcm16ToFloat32(int16Data) {
+  const float32 = new Float32Array(int16Data.length);
+  for (let i = 0; i < int16Data.length; i++) {
+    float32[i] = int16Data[i] / 32768;
+  }
+  return float32;
+}
+function float32ToPcm16(float32Data) {
+  const int16 = new Int16Array(float32Data.length);
+  for (let i = 0; i < float32Data.length; i++) {
+    const s = Math.max(-1, Math.min(1, float32Data[i]));
+    int16[i] = s < 0 ? s * 32768 : s * 32767;
+  }
+  return int16;
+}
+function resample(input, inputRate, outputRate) {
+  if (inputRate === outputRate) {
+    return new Float32Array(input);
+  }
+  const ratio = inputRate / outputRate;
+  const outputLength = Math.round(input.length / ratio);
+  const output = new Float32Array(outputLength);
+  for (let i = 0; i < outputLength; i++) {
+    const pos = i * ratio;
+    const left = Math.floor(pos);
+    const right = Math.min(left + 1, input.length - 1);
+    const weight = pos - left;
+    output[i] = input[left] * (1 - weight) + input[right] * weight;
+  }
+  return output;
+}
+function applyLowPassFilter(data, cutoffFreq, sampleRate) {
+  const dt = 1 / sampleRate;
+  const rc = 1 / (2 * Math.PI * cutoffFreq);
+  const alpha = dt / (rc + dt);
+  const filtered = new Float32Array(data.length);
+  filtered[0] = data[0];
+  for (let i = 1; i < data.length; i++) {
+    filtered[i] = filtered[i - 1] + alpha * (data[i] - filtered[i - 1]);
+  }
+  return filtered;
+}
+function resampleWithAntiAliasing(input, inputRate, outputRate) {
+  if (inputRate === outputRate) {
+    return new Float32Array(input);
+  }
+  let processed = input;
+  if (outputRate < inputRate) {
+    const nyquistFreq = outputRate / 2;
+    const cutoffFreq = nyquistFreq * 0.9;
+    processed = applyLowPassFilter(input, cutoffFreq, inputRate);
+  }
+  return resample(processed, inputRate, outputRate);
+}
+function pcm16ToBytes(data) {
+  return new Uint8Array(data.buffer, data.byteOffset, data.byteLength);
+}
+function bytesToPcm16(bytes) {
+  return new Int16Array(bytes.buffer, bytes.byteOffset, bytes.length / 2);
+}
+function normalizeAudio(data, targetPeak = 0.95) {
+  let maxAbs = 0;
+  for (let i = 0; i < data.length; i++) {
+    maxAbs = Math.max(maxAbs, Math.abs(data[i]));
+  }
+  if (maxAbs === 0) return new Float32Array(data);
+  const scale = targetPeak / maxAbs;
+  const normalized = new Float32Array(data.length);
+  for (let i = 0; i < data.length; i++) {
+    normalized[i] = data[i] * scale;
+  }
+  return normalized;
+}
+function calculateRMS(data) {
+  let sum = 0;
+  let length = data.length;
+  if (data instanceof Uint8Array) {
+    for (let i = 0; i < length; i++) {
+      const v = (data[i] - 128) / 128;
+      sum += v * v;
+    }
+  } else {
+    for (let i = 0; i < length; i++) {
+      sum += data[i] * data[i];
+    }
+  }
+  return Math.sqrt(sum / length);
+}
+var StreamResampler = class {
+  inputBuffer = new Float32Array(0);
+  inputRate;
+  outputRate;
+  constructor(inputRate, outputRate) {
+    this.inputRate = inputRate;
+    this.outputRate = outputRate;
+  }
+  /**
+   * Process a chunk of audio and return resampled data
+   * @param inputChunk Float32Array chunk to process
+   * @param flush If true, output remaining buffered samples
+   * @returns Resampled Float32Array (may be empty if more data needed)
+   */
+  process(inputChunk, flush = false) {
+    const combined = new Float32Array(this.inputBuffer.length + inputChunk.length);
+    combined.set(this.inputBuffer);
+    combined.set(inputChunk, this.inputBuffer.length);
+    const ratio = this.inputRate / this.outputRate;
+    const outputLength = Math.floor(combined.length / ratio);
+    if (outputLength === 0 && !flush) {
+      this.inputBuffer = combined;
+      return new Float32Array(0);
+    }
+    const output = new Float32Array(outputLength);
+    for (let i = 0; i < outputLength; i++) {
+      const pos = i * ratio;
+      const left = Math.floor(pos);
+      const right = Math.min(left + 1, combined.length - 1);
+      const weight = pos - left;
+      output[i] = combined[left] * (1 - weight) + combined[right] * weight;
+    }
+    const remainingSamples = Math.ceil(combined.length - outputLength * ratio);
+    this.inputBuffer = combined.slice(
+      combined.length - remainingSamples
+    );
+    return output;
+  }
+  reset() {
+    this.inputBuffer = new Float32Array(0);
+  }
+};
+
+// src/browser-audio.ts
+var BrowserAudioManager = class {
+  audioContext = null;
+  mediaStreamAudioSourceNode = null;
+  scriptProcessor = null;
+  analyserNode = null;
+  mediaStream = null;
+  // Playback scheduling
+  nextPlaybackTime = 0;
+  activeSources = [];
+  playbackQueue = [];
+  // Configuration
+  inputSampleRate;
+  outputSampleRate;
+  autoGainControl;
+  echoCancellation;
+  noiseSuppression;
+  // Callbacks
+  onAudioInput;
+  onInputError;
+  // Audio processing state
+  isMuted = false;
+  isListening = false;
+  constructor(config = {}) {
+    this.inputSampleRate = config.inputSampleRate ?? AUDIO_CONFIG.SAMPLE_RATE;
+    this.outputSampleRate = config.outputSampleRate ?? AUDIO_CONFIG.SPEAKER_SAMPLE_RATE;
+    this.autoGainControl = config.autoGainControl ?? true;
+    this.echoCancellation = config.echoCancellation ?? true;
+    this.noiseSuppression = config.noiseSuppression ?? true;
+    this.onInputError = config.onInputError;
+  }
+  /**
+   * Initialize the AudioContext and analyser
+   */
+  async init(analyserConfig) {
+    if (this.audioContext) return;
+    const AudioContextClass = window.AudioContext || window.webkitAudioContext;
+    if (!AudioContextClass) {
+      throw new Error("Web Audio API not supported in this browser");
+    }
+    this.audioContext = new AudioContextClass();
+    if (!this.audioContext) {
+      throw new Error("Failed to initialize AudioContext");
+    }
+    if (this.audioContext.state === "suspended") {
+      await this.audioContext.resume();
+      console.log("\u{1F442} AudioContext resumed");
+    }
+    if (analyserConfig?.enabled !== false) {
+      this.analyserNode = this.audioContext.createAnalyser();
+      this.analyserNode.fftSize = analyserConfig?.fftSize ?? 256;
+    }
+  }
+  /**
+   * Start capturing audio from the microphone
+   */
+  async startMicrophone(onAudioInput) {
+    if (!this.audioContext) {
+      await this.init();
+    }
+    try {
+      this.onAudioInput = onAudioInput;
+      this.isListening = true;
+      this.mediaStream = await navigator.mediaDevices.getUserMedia({
+        audio: {
+          autoGainControl: this.autoGainControl,
+          echoCancellation: this.echoCancellation,
+          noiseSuppression: this.noiseSuppression
+        }
+      });
+      this.mediaStreamAudioSourceNode = this.audioContext.createMediaStreamSource(this.mediaStream);
+      const bufferSize = 4096;
+      this.scriptProcessor = this.audioContext.createScriptProcessor(
+        bufferSize,
+        1,
+        // input channels
+        1
+        // output channels
+      );
+      this.mediaStreamAudioSourceNode.connect(this.scriptProcessor);
+      this.scriptProcessor.connect(this.audioContext.destination);
+      if (this.analyserNode) {
+        this.mediaStreamAudioSourceNode.connect(this.analyserNode);
+      }
+      this.scriptProcessor.onaudioprocess = (event) => {
+        this._processAudioInput(event);
+      };
+      console.log("\u{1F3A4} Microphone started");
+    } catch (error) {
+      const err = error instanceof Error ? error : new Error(String(error));
+      if (this.onInputError) this.onInputError(err);
+      throw err;
+    }
+  }
+  /**
+   * Internal method to process microphone audio data
+   */
+  _processAudioInput(event) {
+    if (!this.onAudioInput || !this.audioContext || !this.isListening) return;
+    if (this.isMuted) return;
+    const inputBuffer = event.inputBuffer;
+    const inputData = inputBuffer.getChannelData(0);
+    const outputBuffer = event.outputBuffer;
+    for (let i = 0; i < outputBuffer.getChannelData(0).length; i++) {
+      outputBuffer.getChannelData(0)[i] = 0;
+    }
+    const hardwareRate = this.audioContext.sampleRate;
+    let processedData = new Float32Array(inputData);
+    if (hardwareRate !== this.inputSampleRate) {
+      processedData = resampleWithAntiAliasing(
+        processedData,
+        hardwareRate,
+        this.inputSampleRate
+      );
+    }
+    const int16Data = float32ToPcm16(processedData);
+    const uint8Data = new Uint8Array(
+      int16Data.buffer,
+      int16Data.byteOffset,
+      int16Data.byteLength
+    );
+    this.onAudioInput(uint8Data);
+  }
+  /**
+   * Stop capturing microphone input
+   */
+  stopMicrophone() {
+    this.isListening = false;
+    if (this.mediaStream) {
+      this.mediaStream.getTracks().forEach((track) => track.stop());
+      this.mediaStream = null;
+    }
+    if (this.scriptProcessor) {
+      this.scriptProcessor.disconnect();
+      this.scriptProcessor = null;
+    }
+    if (this.mediaStreamAudioSourceNode) {
+      this.mediaStreamAudioSourceNode.disconnect();
+      this.mediaStreamAudioSourceNode = null;
+    }
+    console.log("\u{1F3A4} Microphone stopped");
+  }
+  /**
+   * Play back audio received from the server
+   * @param pcm16Data Int16 PCM audio data at SPEAKER_SAMPLE_RATE
+   */
+  playAudio(pcm16Data) {
+    if (!this.audioContext) {
+      console.warn("AudioContext not initialized");
+      return;
+    }
+    const int16Array = new Int16Array(
+      pcm16Data.buffer,
+      pcm16Data.byteOffset,
+      pcm16Data.length / 2
+    );
+    const float32Data = pcm16ToFloat32(int16Array);
+    const audioBuffer = this.audioContext.createBuffer(
+      1,
+      float32Data.length,
+      this.outputSampleRate
+    );
+    audioBuffer.getChannelData(0).set(float32Data);
+    this._schedulePlayback(audioBuffer);
+  }
+  /**
+   * Internal method to schedule and play audio with sample-accurate timing
+   */
+  _schedulePlayback(audioBuffer) {
+    if (!this.audioContext) return;
+    const currentTime = this.audioContext.currentTime;
+    const duration = audioBuffer.length / this.outputSampleRate;
+    const startTime = Math.max(
+      currentTime + 0.01,
+      // Minimum 10ms delay
+      this.nextPlaybackTime
+    );
+    this.nextPlaybackTime = startTime + duration;
+    const source = this.audioContext.createBufferSource();
+    source.buffer = audioBuffer;
+    source.connect(this.audioContext.destination);
+    if (this.analyserNode) {
+      source.connect(this.analyserNode);
+    }
+    source.start(startTime);
+    this.activeSources.push(source);
+    source.onended = () => {
+      const index = this.activeSources.indexOf(source);
+      if (index > -1) {
+        this.activeSources.splice(index, 1);
+      }
+    };
+  }
+  /**
+   * Stop all currently playing audio and clear the queue
+   */
+  stopPlayback() {
+    this.activeSources.forEach((source) => {
+      try {
+        source.stop();
+      } catch (e) {
+      }
+    });
+    this.activeSources = [];
+    this.playbackQueue = [];
+    this.nextPlaybackTime = this.audioContext?.currentTime ?? 0;
+    console.log("\u{1F507} Playback stopped");
+  }
+  /**
+   * Toggle mute state
+   */
+  setMuted(muted) {
+    this.isMuted = muted;
+  }
+  /**
+   * Get current mute state
+   */
+  isMicMuted() {
+    return this.isMuted;
+  }
+  /**
+   * Get current amplitude from analyser (for visualization)
+   * Returns value between 0 and 1
+   */
+  getAmplitude() {
+    if (!this.analyserNode) return 0;
+    const dataArray = new Uint8Array(this.analyserNode.frequencyBinCount);
+    this.analyserNode.getByteTimeDomainData(dataArray);
+    const rms = calculateRMS(dataArray);
+    return Math.min(rms * 10, 1);
+  }
+  /**
+   * Get frequency data from analyser for visualization
+   */
+  getFrequencyData() {
+    if (!this.analyserNode) {
+      return new Uint8Array(0);
+    }
+    const dataArray = new Uint8Array(this.analyserNode.frequencyBinCount);
+    this.analyserNode.getByteFrequencyData(dataArray);
+    return dataArray;
+  }
+  /**
+   * Get time-domain data from analyser for waveform visualization
+   */
+  getWaveformData() {
+    if (!this.analyserNode) {
+      return new Uint8Array(0);
+    }
+    const dataArray = new Uint8Array(this.analyserNode.frequencyBinCount);
+    this.analyserNode.getByteTimeDomainData(dataArray);
+    return dataArray;
+  }
+  /**
+   * Cleanup and close AudioContext
+   */
+  cleanup() {
+    this.stopMicrophone();
+    this.stopPlayback();
+    if (this.analyserNode) {
+      this.analyserNode.disconnect();
+      this.analyserNode = null;
+    }
+  }
+  /**
+   * Get current audio context state
+   */
+  getState() {
+    return this.audioContext?.state ?? null;
+  }
+  /**
+   * Check if microphone is currently listening
+   */
+  isRecording() {
+    return this.isListening;
+  }
+};
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   AUDIO_CONFIG,
+  BrowserAudioManager,
   DEFAULT_URLS,
   Language,
+  StreamResampler,
   TTSClient,
   VoiceAgentClient,
   VoiceStyle,
+  applyLowPassFilter,
+  bytesToPcm16,
+  calculateRMS,
+  float32ToPcm16,
+  normalizeAudio,
+  pcm16ToBytes,
+  pcm16ToFloat32,
+  resample,
+  resampleWithAntiAliasing,
   simpleConversation,
   simpleTTS
 });