@tensamin/audio 0.1.1 → 0.1.2

package/dist/index.mjs

@@ -1,356 +1,33 @@
-// src/context/audio-context.ts
-var sharedContext = null;
-var activePipelines = 0;
-function getAudioContext(options) {
-  if (typeof window === "undefined" || typeof AudioContext === "undefined") {
-    throw new Error(
-      "AudioContext is not supported in this environment (browser only)."
-    );
-  }
-  if (!sharedContext || sharedContext.state === "closed") {
-    sharedContext = new AudioContext(options);
-  }
-  return sharedContext;
-}
-function registerPipeline() {
-  activePipelines++;
-}
-function unregisterPipeline() {
-  activePipelines = Math.max(0, activePipelines - 1);
-}
-async function resumeAudioContext() {
-  if (sharedContext && sharedContext.state === "suspended") {
-    await sharedContext.resume();
-  }
-}
-async function suspendAudioContext() {
-  if (sharedContext && sharedContext.state === "running") {
-    await sharedContext.suspend();
-  }
-}
-async function closeAudioContext() {
-  if (sharedContext && sharedContext.state !== "closed") {
-    await sharedContext.close();
-  }
-  sharedContext = null;
-  activePipelines = 0;
-}
-
-// src/pipeline/audio-pipeline.ts
-import mitt from "mitt";
-
-// src/noise-suppression/rnnoise-node.ts
+import "./chunk-WBQAMGXK.mjs";
 import {
-  RnnoiseWorkletNode,
-  loadRnnoise
-} from "@sapphi-red/web-noise-suppressor";
-var DEFAULT_WASM_URL = "https://unpkg.com/@sapphi-red/web-noise-suppressor@0.3.5/dist/rnnoise.wasm";
-var DEFAULT_SIMD_WASM_URL = "https://unpkg.com/@sapphi-red/web-noise-suppressor@0.3.5/dist/rnnoise_simd.wasm";
-var DEFAULT_WORKLET_URL = "https://unpkg.com/@sapphi-red/web-noise-suppressor@0.3.5/dist/noise-suppressor-worklet.min.js";
-var RNNoisePlugin = class {
-  name = "rnnoise-ns";
-  wasmBuffer = null;
-  async createNode(context, config) {
-    if (!config?.enabled) {
-      const pass = context.createGain();
-      return pass;
-    }
-    if (!this.wasmBuffer) {
-      this.wasmBuffer = await loadRnnoise({
-        url: config.wasmUrl || DEFAULT_WASM_URL,
-        simdUrl: DEFAULT_SIMD_WASM_URL
-        // We should probably allow config for this too, but for now default is fine.
-      });
-    }
-    const workletUrl = config.workletUrl || DEFAULT_WORKLET_URL;
-    try {
-      await context.audioWorklet.addModule(workletUrl);
-    } catch (e) {
-      console.warn("Failed to add RNNoise worklet module:", e);
-    }
-    const node = new RnnoiseWorkletNode(context, {
-      wasmBinary: this.wasmBuffer,
-      maxChannels: 1
-      // Mono for now
-    });
-    return node;
-  }
-};
-
-// src/vad/vad-node.ts
-var energyVadWorkletCode = `
-class EnergyVadProcessor extends AudioWorkletProcessor {
-  constructor() {
-    super();
-    this.smoothing = 0.95;
-    this.energy = 0;
-    this.noiseFloor = 0.001;
-  }
-
-  process(inputs, outputs, parameters) {
-    const input = inputs[0];
-    if (!input || !input.length) return true;
-    const channel = input[0];
-    
-    // Calculate RMS
-    let sum = 0;
-    for (let i = 0; i < channel.length; i++) {
-      sum += channel[i] * channel[i];
-    }
-    const rms = Math.sqrt(sum / channel.length);
-
-    // Simple adaptive noise floor (very basic)
-    if (rms < this.noiseFloor) {
-      this.noiseFloor = this.noiseFloor * 0.99 + rms * 0.01;
-    } else {
-      this.noiseFloor = this.noiseFloor * 0.999 + rms * 0.001;
-    }
-
-    // Calculate "probability" based on SNR
-    // This is a heuristic mapping from energy to 0-1
-    const snr = rms / (this.noiseFloor + 1e-6);
-    const probability = Math.min(1, Math.max(0, (snr - 1.5) / 10)); // Arbitrary scaling
-
-    this.port.postMessage({ probability });
-
-    return true;
-  }
-}
-registerProcessor('energy-vad-processor', EnergyVadProcessor);
-`;
-var EnergyVADPlugin = class {
-  name = "energy-vad";
-  async createNode(context, config, onDecision) {
-    const blob = new Blob([energyVadWorkletCode], {
-      type: "application/javascript"
-    });
-    const url = URL.createObjectURL(blob);
-    try {
-      await context.audioWorklet.addModule(url);
-    } catch (e) {
-      console.warn("Failed to add Energy VAD worklet:", e);
-      throw e;
-    } finally {
-      URL.revokeObjectURL(url);
-    }
-    const node = new AudioWorkletNode(context, "energy-vad-processor");
-    node.port.onmessage = (event) => {
-      const { probability } = event.data;
-      onDecision(probability);
-    };
-    return node;
-  }
-};
-
-// src/extensibility/plugins.ts
-var nsPlugins = /* @__PURE__ */ new Map();
-var vadPlugins = /* @__PURE__ */ new Map();
-var defaultNs = new RNNoisePlugin();
-nsPlugins.set(defaultNs.name, defaultNs);
-var defaultVad = new EnergyVADPlugin();
-vadPlugins.set(defaultVad.name, defaultVad);
-function registerNoiseSuppressionPlugin(plugin) {
-  nsPlugins.set(plugin.name, plugin);
-}
-function registerVADPlugin(plugin) {
-  vadPlugins.set(plugin.name, plugin);
-}
-function getNoiseSuppressionPlugin(name) {
-  if (!name) return defaultNs;
-  const plugin = nsPlugins.get(name);
-  if (!plugin) {
-    console.warn(
-      `Noise suppression plugin '${name}' not found, falling back to default.`
-    );
-    return defaultNs;
-  }
-  return plugin;
-}
-function getVADPlugin(name) {
-  if (!name) return defaultVad;
-  const plugin = vadPlugins.get(name);
-  if (!plugin) {
-    console.warn(`VAD plugin '${name}' not found, falling back to default.`);
-    return defaultVad;
-  }
-  return plugin;
-}
-
-// src/vad/vad-state.ts
-var VADStateMachine = class {
-  config;
-  currentState = "silent";
-  lastSpeechTime = 0;
-  speechStartTime = 0;
-  frameDurationMs = 20;
-  // Assumed frame duration, updated by calls
-  constructor(config) {
-    this.config = {
-      enabled: config?.enabled ?? true,
-      pluginName: config?.pluginName ?? "energy-vad",
-      startThreshold: config?.startThreshold ?? 0.5,
-      stopThreshold: config?.stopThreshold ?? 0.4,
-      hangoverMs: config?.hangoverMs ?? 300,
-      preRollMs: config?.preRollMs ?? 200
-    };
-  }
-  updateConfig(config) {
-    this.config = { ...this.config, ...config };
-  }
-  processFrame(probability, timestamp) {
-    const { startThreshold, stopThreshold, hangoverMs } = this.config;
-    let newState = this.currentState;
-    if (this.currentState === "silent" || this.currentState === "speech_ending") {
-      if (probability >= startThreshold) {
-        newState = "speech_starting";
-        this.speechStartTime = timestamp;
-        this.lastSpeechTime = timestamp;
-      } else {
-        newState = "silent";
-      }
-    } else if (this.currentState === "speech_starting" || this.currentState === "speaking") {
-      if (probability >= stopThreshold) {
-        newState = "speaking";
-        this.lastSpeechTime = timestamp;
-      } else {
-        const timeSinceSpeech = timestamp - this.lastSpeechTime;
-        if (timeSinceSpeech < hangoverMs) {
-          newState = "speaking";
-        } else {
-          newState = "speech_ending";
-        }
-      }
-    }
-    if (newState === "speech_starting") newState = "speaking";
-    if (newState === "speech_ending") newState = "silent";
-    this.currentState = newState;
-    return {
-      isSpeaking: newState === "speaking",
-      probability,
-      state: newState
-    };
-  }
-};
-
-// src/pipeline/audio-pipeline.ts
-async function createAudioPipeline(sourceTrack, config = {}) {
-  const context = getAudioContext();
-  registerPipeline();
-  const fullConfig = {
-    noiseSuppression: { enabled: true, ...config.noiseSuppression },
-    vad: { enabled: true, ...config.vad },
-    output: {
-      speechGain: 1,
-      silenceGain: 0,
-      gainRampTime: 0.02,
-      ...config.output
-    },
-    livekit: { manageTrackMute: false, ...config.livekit }
-  };
-  const sourceStream = new MediaStream([sourceTrack]);
-  const sourceNode = context.createMediaStreamSource(sourceStream);
-  const nsPlugin = getNoiseSuppressionPlugin(
-    fullConfig.noiseSuppression?.pluginName
-  );
-  const nsNode = await nsPlugin.createNode(
-    context,
-    fullConfig.noiseSuppression
-  );
-  const vadPlugin = getVADPlugin(fullConfig.vad?.pluginName);
-  const vadStateMachine = new VADStateMachine(fullConfig.vad);
-  const emitter = mitt();
-  const vadNode = await vadPlugin.createNode(
-    context,
-    fullConfig.vad,
-    (prob) => {
-      const timestamp = context.currentTime * 1e3;
-      const newState = vadStateMachine.processFrame(prob, timestamp);
-      if (newState.state !== lastVadState.state || Math.abs(newState.probability - lastVadState.probability) > 0.1) {
-        emitter.emit("vadChange", newState);
-        lastVadState = newState;
-        updateGain(newState);
-      }
-    }
-  );
-  let lastVadState = {
-    isSpeaking: false,
-    probability: 0,
-    state: "silent"
-  };
-  const splitter = context.createGain();
-  sourceNode.connect(nsNode);
-  nsNode.connect(splitter);
-  splitter.connect(vadNode);
-  const delayNode = context.createDelay(1);
-  const preRollSeconds = (fullConfig.vad?.preRollMs ?? 200) / 1e3;
-  delayNode.delayTime.value = preRollSeconds;
-  const gainNode = context.createGain();
-  gainNode.gain.value = fullConfig.output?.silenceGain ?? 0;
-  const destination = context.createMediaStreamDestination();
-  splitter.connect(delayNode);
-  delayNode.connect(gainNode);
-  gainNode.connect(destination);
-  function updateGain(state) {
-    const { speechGain, silenceGain, gainRampTime } = fullConfig.output;
-    const targetGain = state.isSpeaking ? speechGain ?? 1 : silenceGain ?? 0;
-    const now = context.currentTime;
-    gainNode.gain.setTargetAtTime(targetGain, now, gainRampTime ?? 0.02);
-  }
-  function dispose() {
-    sourceNode.disconnect();
-    nsNode.disconnect();
-    splitter.disconnect();
-    vadNode.disconnect();
-    delayNode.disconnect();
-    gainNode.disconnect();
-    destination.stream.getTracks().forEach((t) => t.stop());
-    unregisterPipeline();
-  }
-  return {
-    processedTrack: destination.stream.getAudioTracks()[0],
-    events: emitter,
-    get state() {
-      return lastVadState;
-    },
-    setConfig: (newConfig) => {
-      if (newConfig.vad) {
-        vadStateMachine.updateConfig(newConfig.vad);
-      }
-    },
-    dispose
-  };
-}
-
-// src/livekit/integration.ts
-async function attachProcessingToTrack(track, config = {}) {
-  const originalTrack = track.mediaStreamTrack;
-  const pipeline = await createAudioPipeline(originalTrack, config);
-  await track.replaceTrack(pipeline.processedTrack);
-  if (config.livekit?.manageTrackMute) {
-    let isVadMuted = false;
-    pipeline.events.on("vadChange", async (state) => {
-      if (state.isSpeaking) {
-        if (isVadMuted) {
-          await track.unmute();
-          isVadMuted = false;
-        }
-      } else {
-        if (!track.isMuted) {
-          await track.mute();
-          isVadMuted = true;
-        }
-      }
-    });
-  }
-  const originalDispose = pipeline.dispose;
-  pipeline.dispose = () => {
-    if (originalTrack.readyState === "live") {
-      track.replaceTrack(originalTrack).catch(console.error);
-    }
-    originalDispose();
-  };
-  return pipeline;
-}
+  attachProcessingToTrack
+} from "./chunk-HFSKQ33X.mjs";
+import {
+  createAudioPipeline
+} from "./chunk-QU7E5HBA.mjs";
+import {
+  VADStateMachine
+} from "./chunk-JJASCVEW.mjs";
+import {
+  closeAudioContext,
+  getAudioContext,
+  registerPipeline,
+  resumeAudioContext,
+  suspendAudioContext,
+  unregisterPipeline
+} from "./chunk-OZ7KMC4S.mjs";
+import {
+  getNoiseSuppressionPlugin,
+  getVADPlugin,
+  registerNoiseSuppressionPlugin,
+  registerVADPlugin
+} from "./chunk-FS635GMR.mjs";
+import {
+  RNNoisePlugin
+} from "./chunk-SDTOKWM2.mjs";
+import {
+  EnergyVADPlugin
+} from "./chunk-UMU2KIB6.mjs";
 export {
   EnergyVADPlugin,
   RNNoisePlugin,

package/dist/livekit/integration.d.mts

@@ -0,0 +1,11 @@
+import { LocalAudioTrack } from 'livekit-client';
+import { AudioProcessingConfig, AudioPipelineHandle } from '../types.mjs';
+import 'mitt';
+
+/**
+ * Attaches the audio processing pipeline to a LiveKit LocalAudioTrack.
+ * This replaces the underlying MediaStreamTrack with the processed one.
+ */
+declare function attachProcessingToTrack(track: LocalAudioTrack, config?: AudioProcessingConfig): Promise<AudioPipelineHandle>;
+
+export { attachProcessingToTrack };

package/dist/livekit/integration.d.ts

@@ -0,0 +1,11 @@
+import { LocalAudioTrack } from 'livekit-client';
+import { AudioProcessingConfig, AudioPipelineHandle } from '../types.js';
+import 'mitt';
+
+/**
+ * Attaches the audio processing pipeline to a LiveKit LocalAudioTrack.
+ * This replaces the underlying MediaStreamTrack with the processed one.
+ */
+declare function attachProcessingToTrack(track: LocalAudioTrack, config?: AudioProcessingConfig): Promise<AudioPipelineHandle>;
+
+export { attachProcessingToTrack };