@livekit/agents 0.0.0-20260120144724

package/dist/vad.cjs

@@ -0,0 +1,211 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var vad_exports = {};
+__export(vad_exports, {
+  VAD: () => VAD,
+  VADEventType: () => VADEventType,
+  VADStream: () => VADStream
+});
+module.exports = __toCommonJS(vad_exports);
+var import_node_events = require("node:events");
+var import_log = require("./log.cjs");
+var import_deferred_stream = require("./stream/deferred_stream.cjs");
+var import_identity_transform = require("./stream/identity_transform.cjs");
+var VADEventType = /* @__PURE__ */ ((VADEventType2) => {
+  VADEventType2[VADEventType2["START_OF_SPEECH"] = 0] = "START_OF_SPEECH";
+  VADEventType2[VADEventType2["INFERENCE_DONE"] = 1] = "INFERENCE_DONE";
+  VADEventType2[VADEventType2["END_OF_SPEECH"] = 2] = "END_OF_SPEECH";
+  VADEventType2[VADEventType2["METRICS_COLLECTED"] = 3] = "METRICS_COLLECTED";
+  return VADEventType2;
+})(VADEventType || {});
+class VAD extends import_node_events.EventEmitter {
+  #capabilities;
+  constructor(capabilities) {
+    super();
+    this.#capabilities = capabilities;
+  }
+  get capabilities() {
+    return this.#capabilities;
+  }
+  async close() {
+    return;
+  }
+}
+class VADStream {
+  static FLUSH_SENTINEL = Symbol("FLUSH_SENTINEL");
+  input = new import_identity_transform.IdentityTransform();
+  output = new import_identity_transform.IdentityTransform();
+  inputWriter;
+  inputReader;
+  outputWriter;
+  outputReader;
+  closed = false;
+  inputClosed = false;
+  #vad;
+  #lastActivityTime = BigInt(0);
+  logger = (0, import_log.log)();
+  deferredInputStream;
+  metricsStream;
+  constructor(vad) {
+    this.#vad = vad;
+    this.deferredInputStream = new import_deferred_stream.DeferredReadableStream();
+    this.inputWriter = this.input.writable.getWriter();
+    this.inputReader = this.input.readable.getReader();
+    this.outputWriter = this.output.writable.getWriter();
+    const [outputStream, metricsStream] = this.output.readable.tee();
+    this.metricsStream = metricsStream;
+    this.outputReader = outputStream.getReader();
+    this.pumpDeferredStream();
+    this.monitorMetrics();
+  }
+  /**
+   * Reads from the deferred input stream and forwards chunks to the input writer.
+   *
+   * Note: we can't just do this.deferredInputStream.stream.pipeTo(this.input.writable)
+   * because the inputWriter locks the this.input.writable stream. All writes must go through
+   * the inputWriter.
+   */
+  async pumpDeferredStream() {
+    const reader = this.deferredInputStream.stream.getReader();
+    try {
+      while (true) {
+        const { done, value } = await reader.read();
+        if (done) break;
+        await this.inputWriter.write(value);
+      }
+    } catch (e) {
+      this.logger.error(`Error pumping deferred stream: ${e}`);
+      throw e;
+    } finally {
+      reader.releaseLock();
+    }
+  }
+  async monitorMetrics() {
+    let inferenceDurationTotalMs = 0;
+    let inferenceCount = 0;
+    const metricsReader = this.metricsStream.getReader();
+    while (true) {
+      const { done, value } = await metricsReader.read();
+      if (done) {
+        break;
+      }
+      switch (value.type) {
+        case 0 /* START_OF_SPEECH */:
+          inferenceCount++;
+          if (inferenceCount >= 1 / this.#vad.capabilities.updateInterval) {
+            this.#vad.emit("metrics_collected", {
+              type: "vad_metrics",
+              timestamp: Date.now(),
+              idleTimeMs: Math.trunc(
+                Number((process.hrtime.bigint() - this.#lastActivityTime) / BigInt(1e6))
+              ),
+              inferenceDurationTotalMs,
+              inferenceCount,
+              label: this.#vad.label
+            });
+            inferenceCount = 0;
+            inferenceDurationTotalMs = 0;
+          }
+          break;
+        case 1 /* INFERENCE_DONE */:
+          inferenceDurationTotalMs += Math.round(value.inferenceDuration);
+          this.#lastActivityTime = process.hrtime.bigint();
+          break;
+        case 2 /* END_OF_SPEECH */:
+          this.#lastActivityTime = process.hrtime.bigint();
+          break;
+      }
+    }
+  }
+  /**
+   * Safely send a VAD event to the output stream, handling writer release errors during shutdown.
+   * @returns true if the event was sent, false if the stream is closing
+   * @throws Error if an unexpected error occurs
+   */
+  sendVADEvent(event) {
+    if (this.closed) {
+      return false;
+    }
+    try {
+      this.outputWriter.write(event);
+      return true;
+    } catch (e) {
+      throw e;
+    }
+  }
+  updateInputStream(audioStream) {
+    this.deferredInputStream.setSource(audioStream);
+  }
+  detachInputStream() {
+    this.deferredInputStream.detachSource();
+  }
+  /** @deprecated Use `updateInputStream` instead */
+  pushFrame(frame) {
+    if (this.inputClosed) {
+      throw new Error("Input is closed");
+    }
+    if (this.closed) {
+      throw new Error("Stream is closed");
+    }
+    this.inputWriter.write(frame);
+  }
+  flush() {
+    if (this.inputClosed) {
+      throw new Error("Input is closed");
+    }
+    if (this.closed) {
+      throw new Error("Stream is closed");
+    }
+    this.inputWriter.write(VADStream.FLUSH_SENTINEL);
+  }
+  endInput() {
+    if (this.inputClosed) {
+      throw new Error("Input is closed");
+    }
+    if (this.closed) {
+      throw new Error("Stream is closed");
+    }
+    this.inputClosed = true;
+    this.input.writable.close();
+  }
+  async next() {
+    return this.outputReader.read().then(({ done, value }) => {
+      if (done) {
+        return { done: true, value: void 0 };
+      }
+      return { done: false, value };
+    });
+  }
+  close() {
+    this.outputWriter.releaseLock();
+    this.outputReader.cancel();
+    this.output.writable.close();
+    this.closed = true;
+  }
+  [Symbol.asyncIterator]() {
+    return this;
+  }
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  VAD,
+  VADEventType,
+  VADStream
+});
+//# sourceMappingURL=vad.cjs.map

package/dist/vad.cjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/vad.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport type { TypedEventEmitter as TypedEmitter } from '@livekit/typed-emitter';\nimport { EventEmitter } from 'node:events';\nimport type {\n  ReadableStream,\n  ReadableStreamDefaultReader,\n  WritableStreamDefaultWriter,\n} from 'node:stream/web';\nimport { log } from './log.js';\nimport type { VADMetrics } from './metrics/base.js';\nimport { DeferredReadableStream } from './stream/deferred_stream.js';\nimport { IdentityTransform } from './stream/identity_transform.js';\n\nexport enum VADEventType {\n  START_OF_SPEECH,\n  INFERENCE_DONE,\n  END_OF_SPEECH,\n  METRICS_COLLECTED,\n}\n\nexport interface VADEvent {\n  /** Type of the VAD event (e.g., start of speech, end of speech, inference done). */\n  type: VADEventType;\n  /**\n   * Index of the audio sample where the event occurred, relative to the inference sample rate.\n   */\n  samplesIndex: number;\n  /** Timestamp when the event was fired. */\n  timestamp: number;\n  /** Duration of the speech segment. */\n  speechDuration: number;\n  /** Duration of the silence segment. */\n  silenceDuration: number;\n  /**\n   * List of audio frames associated with the speech.\n   *\n   * @remarks\n   * - For `start_of_speech` events, this contains the audio chunks that triggered the detection.\n   * - For `inference_done` events, this contains the audio chunks that were processed.\n   * - For `end_of_speech` events, this contains the complete user speech.\n   */\n  frames: AudioFrame[];\n  /** Probability that speech is present (only for `INFERENCE_DONE` events). */\n  probability: number;\n  /** Time taken to perform the inference, in seconds (only for `INFERENCE_DONE` events). */\n  inferenceDuration: number;\n  /** Indicates whether speech was detected in the frames. */\n  speaking: boolean;\n  /** Threshold used to detect silence. */\n  rawAccumulatedSilence: number;\n  /** Threshold used to detect speech. */\n  rawAccumulatedSpeech: number;\n}\n\nexport interface VADCapabilities {\n  updateInterval: number;\n}\n\nexport type VADCallbacks = {\n  ['metrics_collected']: (metrics: VADMetrics) => void;\n};\n\nexport abstract class VAD extends (EventEmitter as new () => TypedEmitter<VADCallbacks>) {\n  #capabilities: VADCapabilities;\n  abstract label: string;\n\n  constructor(capabilities: VADCapabilities) {\n    super();\n    this.#capabilities = capabilities;\n  }\n\n  get capabilities(): VADCapabilities {\n    return this.#capabilities;\n  }\n\n  /**\n   * Returns a {@link VADStream} that can be used to push audio frames and receive VAD events.\n   */\n  abstract stream(): VADStream;\n\n  async close(): Promise<void> {\n    return;\n  }\n}\n\nexport abstract class VADStream implements AsyncIterableIterator<VADEvent> {\n  protected static readonly FLUSH_SENTINEL = Symbol('FLUSH_SENTINEL');\n  protected input = new IdentityTransform<AudioFrame | typeof VADStream.FLUSH_SENTINEL>();\n  protected output = new IdentityTransform<VADEvent>();\n  protected inputWriter: WritableStreamDefaultWriter<AudioFrame | typeof VADStream.FLUSH_SENTINEL>;\n  protected inputReader: ReadableStreamDefaultReader<AudioFrame | typeof VADStream.FLUSH_SENTINEL>;\n  protected outputWriter: WritableStreamDefaultWriter<VADEvent>;\n  protected outputReader: ReadableStreamDefaultReader<VADEvent>;\n  protected closed = false;\n  protected inputClosed = false;\n\n  #vad: VAD;\n  #lastActivityTime = BigInt(0);\n  private logger = log();\n  private deferredInputStream: DeferredReadableStream<AudioFrame>;\n\n  private metricsStream: ReadableStream<VADEvent>;\n  constructor(vad: VAD) {\n    this.#vad = vad;\n    this.deferredInputStream = new DeferredReadableStream<AudioFrame>();\n\n    this.inputWriter = this.input.writable.getWriter();\n    this.inputReader = this.input.readable.getReader();\n    this.outputWriter = this.output.writable.getWriter();\n\n    const [outputStream, metricsStream] = this.output.readable.tee();\n    this.metricsStream = metricsStream;\n    this.outputReader = outputStream.getReader();\n\n    this.pumpDeferredStream();\n    this.monitorMetrics();\n  }\n\n  /**\n   * Reads from the deferred input stream and forwards chunks to the input writer.\n   *\n   * Note: we can't just do this.deferredInputStream.stream.pipeTo(this.input.writable)\n   * because the inputWriter locks the this.input.writable stream. All writes must go through\n   * the inputWriter.\n   */\n  private async pumpDeferredStream() {\n    const reader = this.deferredInputStream.stream.getReader();\n    try {\n      while (true) {\n        const { done, value } = await reader.read();\n        if (done) break;\n        await this.inputWriter.write(value);\n      }\n    } catch (e) {\n      this.logger.error(`Error pumping deferred stream: ${e}`);\n      throw e;\n    } finally {\n      reader.releaseLock();\n    }\n  }\n\n  protected async monitorMetrics() {\n    let inferenceDurationTotalMs = 0;\n    let inferenceCount = 0;\n    const metricsReader = this.metricsStream.getReader();\n    while (true) {\n      const { done, value } = await metricsReader.read();\n      if (done) {\n        break;\n      }\n      switch (value.type) {\n        case VADEventType.START_OF_SPEECH:\n          inferenceCount++;\n          if (inferenceCount >= 1 / this.#vad.capabilities.updateInterval) {\n            this.#vad.emit('metrics_collected', {\n              type: 'vad_metrics',\n              timestamp: Date.now(),\n              idleTimeMs: Math.trunc(\n                Number((process.hrtime.bigint() - this.#lastActivityTime) / BigInt(1000000)),\n              ),\n              inferenceDurationTotalMs,\n              inferenceCount,\n              label: this.#vad.label,\n            });\n\n            inferenceCount = 0;\n            inferenceDurationTotalMs = 0;\n          }\n          break;\n        case VADEventType.INFERENCE_DONE:\n          inferenceDurationTotalMs += Math.round(value.inferenceDuration);\n          this.#lastActivityTime = process.hrtime.bigint();\n          break;\n        case VADEventType.END_OF_SPEECH:\n          this.#lastActivityTime = process.hrtime.bigint();\n          break;\n      }\n    }\n  }\n\n  /**\n   * Safely send a VAD event to the output stream, handling writer release errors during shutdown.\n   * @returns true if the event was sent, false if the stream is closing\n   * @throws Error if an unexpected error occurs\n   */\n  protected sendVADEvent(event: VADEvent): boolean {\n    if (this.closed) {\n      return false;\n    }\n\n    try {\n      this.outputWriter.write(event);\n      return true;\n    } catch (e) {\n      throw e;\n    }\n  }\n\n  updateInputStream(audioStream: ReadableStream<AudioFrame>) {\n    this.deferredInputStream.setSource(audioStream);\n  }\n\n  detachInputStream() {\n    this.deferredInputStream.detachSource();\n  }\n\n  /** @deprecated Use `updateInputStream` instead */\n  pushFrame(frame: AudioFrame) {\n    // TODO(AJS-395): remove this method\n    if (this.inputClosed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.inputWriter.write(frame);\n  }\n\n  flush() {\n    if (this.inputClosed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.inputWriter.write(VADStream.FLUSH_SENTINEL);\n  }\n\n  endInput() {\n    if (this.inputClosed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.inputClosed = true;\n    this.input.writable.close();\n  }\n\n  async next(): Promise<IteratorResult<VADEvent>> {\n    return this.outputReader.read().then(({ done, value }) => {\n      if (done) {\n        return { done: true, value: undefined };\n      }\n      return { done: false, value };\n    });\n  }\n\n  close() {\n    this.outputWriter.releaseLock();\n    this.outputReader.cancel();\n    this.output.writable.close();\n    this.closed = true;\n  }\n\n  [Symbol.asyncIterator](): VADStream {\n    return this;\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAKA,yBAA6B;AAM7B,iBAAoB;AAEpB,6BAAuC;AACvC,gCAAkC;AAE3B,IAAK,eAAL,kBAAKA,kBAAL;AACL,EAAAA,4BAAA;AACA,EAAAA,4BAAA;AACA,EAAAA,4BAAA;AACA,EAAAA,4BAAA;AAJU,SAAAA;AAAA,GAAA;AAiDL,MAAe,YAAa,gCAAsD;AAAA,EACvF;AAAA,EAGA,YAAY,cAA+B;AACzC,UAAM;AACN,SAAK,gBAAgB;AAAA,EACvB;AAAA,EAEA,IAAI,eAAgC;AAClC,WAAO,KAAK;AAAA,EACd;AAAA,EAOA,MAAM,QAAuB;AAC3B;AAAA,EACF;AACF;AAEO,MAAe,UAAqD;AAAA,EACzE,OAA0B,iBAAiB,OAAO,gBAAgB;AAAA,EACxD,QAAQ,IAAI,4CAAgE;AAAA,EAC5E,SAAS,IAAI,4CAA4B;AAAA,EACzC;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA,SAAS;AAAA,EACT,cAAc;AAAA,EAExB;AAAA,EACA,oBAAoB,OAAO,CAAC;AAAA,EACpB,aAAS,gBAAI;AAAA,EACb;AAAA,EAEA;AAAA,EACR,YAAY,KAAU;AACpB,SAAK,OAAO;AACZ,SAAK,sBAAsB,IAAI,8CAAmC;AAElE,SAAK,cAAc,KAAK,MAAM,SAAS,UAAU;AACjD,SAAK,cAAc,KAAK,MAAM,SAAS,UAAU;AACjD,SAAK,eAAe,KAAK,OAAO,SAAS,UAAU;AAEnD,UAAM,CAAC,cAAc,aAAa,IAAI,KAAK,OAAO,SAAS,IAAI;AAC/D,SAAK,gBAAgB;AACrB,SAAK,eAAe,aAAa,UAAU;AAE3C,SAAK,mBAAmB;AACxB,SAAK,eAAe;AAAA,EACtB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EASA,MAAc,qBAAqB;AACjC,UAAM,SAAS,KAAK,oBAAoB,OAAO,UAAU;AACzD,QAAI;AACF,aAAO,MAAM;AACX,cAAM,EAAE,MAAM,MAAM,IAAI,MAAM,OAAO,KAAK;AAC1C,YAAI,KAAM;AACV,cAAM,KAAK,YAAY,MAAM,KAAK;AAAA,MACpC;AAAA,IACF,SAAS,GAAG;AACV,WAAK,OAAO,MAAM,kCAAkC,CAAC,EAAE;AACvD,YAAM;AAAA,IACR,UAAE;AACA,aAAO,YAAY;AAAA,IACrB;AAAA,EACF;AAAA,EAEA,MAAgB,iBAAiB;AAC/B,QAAI,2BAA2B;AAC/B,QAAI,iBAAiB;AACrB,UAAM,gBAAgB,KAAK,cAAc,UAAU;AACnD,WAAO,MAAM;AACX,YAAM,EAAE,MAAM,MAAM,IAAI,MAAM,cAAc,KAAK;AACjD,UAAI,MAAM;AACR;AAAA,MACF;AACA,cAAQ,MAAM,MAAM;AAAA,QAClB,KAAK;AACH;AACA,cAAI,kBAAkB,IAAI,KAAK,KAAK,aAAa,gBAAgB;AAC/D,iBAAK,KAAK,KAAK,qBAAqB;AAAA,cAClC,MAAM;AAAA,cACN,WAAW,KAAK,IAAI;AAAA,cACpB,YAAY,KAAK;AAAA,gBACf,QAAQ,QAAQ,OAAO,OAAO,IAAI,KAAK,qBAAqB,OAAO,GAAO,CAAC;AAAA,cAC7E;AAAA,cACA;AAAA,cACA;AAAA,cACA,OAAO,KAAK,KAAK;AAAA,YACnB,CAAC;AAED,6BAAiB;AACjB,uCAA2B;AAAA,UAC7B;AACA;AAAA,QACF,KAAK;AACH,sCAA4B,KAAK,MAAM,MAAM,iBAAiB;AAC9D,eAAK,oBAAoB,QAAQ,OAAO,OAAO;AAC/C;AAAA,QACF,KAAK;AACH,eAAK,oBAAoB,QAAQ,OAAO,OAAO;AAC/C;AAAA,MACJ;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOU,aAAa,OAA0B;AAC/C,QAAI,KAAK,QAAQ;AACf,aAAO;AAAA,IACT;AAEA,QAAI;AACF,WAAK,aAAa,MAAM,KAAK;AAC7B,aAAO;AAAA,IACT,SAAS,GAAG;AACV,YAAM;AAAA,IACR;AAAA,EACF;AAAA,EAEA,kBAAkB,aAAyC;AACzD,SAAK,oBAAoB,UAAU,WAAW;AAAA,EAChD;AAAA,EAEA,oBAAoB;AAClB,SAAK,oBAAoB,aAAa;AAAA,EACxC;AAAA;AAAA,EAGA,UAAU,OAAmB;AAE3B,QAAI,KAAK,aAAa;AACpB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,YAAY,MAAM,KAAK;AAAA,EAC9B;AAAA,EAEA,QAAQ;AACN,QAAI,KAAK,aAAa;AACpB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,YAAY,MAAM,UAAU,cAAc;AAAA,EACjD;AAAA,EAEA,WAAW;AACT,QAAI,KAAK,aAAa;AACpB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,cAAc;AACnB,SAAK,MAAM,SAAS,MAAM;AAAA,EAC5B;AAAA,EAEA,MAAM,OAA0C;AAC9C,WAAO,KAAK,aAAa,KAAK,EAAE,KAAK,CAAC,EAAE,MAAM,MAAM,MAAM;AACxD,UAAI,MAAM;AACR,eAAO,EAAE,MAAM,MAAM,OAAO,OAAU;AAAA,MACxC;AACA,aAAO,EAAE,MAAM,OAAO,MAAM;AAAA,IAC9B,CAAC;AAAA,EACH;AAAA,EAEA,QAAQ;AACN,SAAK,aAAa,YAAY;AAC9B,SAAK,aAAa,OAAO;AACzB,SAAK,OAAO,SAAS,MAAM;AAC3B,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,CAAC,OAAO,aAAa,IAAe;AAClC,WAAO;AAAA,EACT;AACF;","names":["VADEventType"]}

package/dist/vad.d.cts

@@ -0,0 +1,105 @@
+/// <reference types="node" resolution-mode="require"/>
+import type { AudioFrame } from '@livekit/rtc-node';
+import type { TypedEventEmitter as TypedEmitter } from '@livekit/typed-emitter';
+import type { ReadableStream, ReadableStreamDefaultReader, WritableStreamDefaultWriter } from 'node:stream/web';
+import type { VADMetrics } from './metrics/base.js';
+import { IdentityTransform } from './stream/identity_transform.js';
+export declare enum VADEventType {
+    START_OF_SPEECH = 0,
+    INFERENCE_DONE = 1,
+    END_OF_SPEECH = 2,
+    METRICS_COLLECTED = 3
+}
+export interface VADEvent {
+    /** Type of the VAD event (e.g., start of speech, end of speech, inference done). */
+    type: VADEventType;
+    /**
+     * Index of the audio sample where the event occurred, relative to the inference sample rate.
+     */
+    samplesIndex: number;
+    /** Timestamp when the event was fired. */
+    timestamp: number;
+    /** Duration of the speech segment. */
+    speechDuration: number;
+    /** Duration of the silence segment. */
+    silenceDuration: number;
+    /**
+     * List of audio frames associated with the speech.
+     *
+     * @remarks
+     * - For `start_of_speech` events, this contains the audio chunks that triggered the detection.
+     * - For `inference_done` events, this contains the audio chunks that were processed.
+     * - For `end_of_speech` events, this contains the complete user speech.
+     */
+    frames: AudioFrame[];
+    /** Probability that speech is present (only for `INFERENCE_DONE` events). */
+    probability: number;
+    /** Time taken to perform the inference, in seconds (only for `INFERENCE_DONE` events). */
+    inferenceDuration: number;
+    /** Indicates whether speech was detected in the frames. */
+    speaking: boolean;
+    /** Threshold used to detect silence. */
+    rawAccumulatedSilence: number;
+    /** Threshold used to detect speech. */
+    rawAccumulatedSpeech: number;
+}
+export interface VADCapabilities {
+    updateInterval: number;
+}
+export type VADCallbacks = {
+    ['metrics_collected']: (metrics: VADMetrics) => void;
+};
+declare const VAD_base: new () => TypedEmitter<VADCallbacks>;
+export declare abstract class VAD extends VAD_base {
+    #private;
+    abstract label: string;
+    constructor(capabilities: VADCapabilities);
+    get capabilities(): VADCapabilities;
+    /**
+     * Returns a {@link VADStream} that can be used to push audio frames and receive VAD events.
+     */
+    abstract stream(): VADStream;
+    close(): Promise<void>;
+}
+export declare abstract class VADStream implements AsyncIterableIterator<VADEvent> {
+    #private;
+    protected static readonly FLUSH_SENTINEL: unique symbol;
+    protected input: IdentityTransform<AudioFrame | typeof VADStream.FLUSH_SENTINEL>;
+    protected output: IdentityTransform<VADEvent>;
+    protected inputWriter: WritableStreamDefaultWriter<AudioFrame | typeof VADStream.FLUSH_SENTINEL>;
+    protected inputReader: ReadableStreamDefaultReader<AudioFrame | typeof VADStream.FLUSH_SENTINEL>;
+    protected outputWriter: WritableStreamDefaultWriter<VADEvent>;
+    protected outputReader: ReadableStreamDefaultReader<VADEvent>;
+    protected closed: boolean;
+    protected inputClosed: boolean;
+    private logger;
+    private deferredInputStream;
+    private metricsStream;
+    constructor(vad: VAD);
+    /**
+     * Reads from the deferred input stream and forwards chunks to the input writer.
+     *
+     * Note: we can't just do this.deferredInputStream.stream.pipeTo(this.input.writable)
+     * because the inputWriter locks the this.input.writable stream. All writes must go through
+     * the inputWriter.
+     */
+    private pumpDeferredStream;
+    protected monitorMetrics(): Promise<void>;
+    /**
+     * Safely send a VAD event to the output stream, handling writer release errors during shutdown.
+     * @returns true if the event was sent, false if the stream is closing
+     * @throws Error if an unexpected error occurs
+     */
+    protected sendVADEvent(event: VADEvent): boolean;
+    updateInputStream(audioStream: ReadableStream<AudioFrame>): void;
+    detachInputStream(): void;
+    /** @deprecated Use `updateInputStream` instead */
+    pushFrame(frame: AudioFrame): void;
+    flush(): void;
+    endInput(): void;
+    next(): Promise<IteratorResult<VADEvent>>;
+    close(): void;
+    [Symbol.asyncIterator](): VADStream;
+}
+export {};
+//# sourceMappingURL=vad.d.ts.map

package/dist/vad.d.ts

@@ -0,0 +1,105 @@
+/// <reference types="node" resolution-mode="require"/>
+import type { AudioFrame } from '@livekit/rtc-node';
+import type { TypedEventEmitter as TypedEmitter } from '@livekit/typed-emitter';
+import type { ReadableStream, ReadableStreamDefaultReader, WritableStreamDefaultWriter } from 'node:stream/web';
+import type { VADMetrics } from './metrics/base.js';
+import { IdentityTransform } from './stream/identity_transform.js';
+export declare enum VADEventType {
+    START_OF_SPEECH = 0,
+    INFERENCE_DONE = 1,
+    END_OF_SPEECH = 2,
+    METRICS_COLLECTED = 3
+}
+export interface VADEvent {
+    /** Type of the VAD event (e.g., start of speech, end of speech, inference done). */
+    type: VADEventType;
+    /**
+     * Index of the audio sample where the event occurred, relative to the inference sample rate.
+     */
+    samplesIndex: number;
+    /** Timestamp when the event was fired. */
+    timestamp: number;
+    /** Duration of the speech segment. */
+    speechDuration: number;
+    /** Duration of the silence segment. */
+    silenceDuration: number;
+    /**
+     * List of audio frames associated with the speech.
+     *
+     * @remarks
+     * - For `start_of_speech` events, this contains the audio chunks that triggered the detection.
+     * - For `inference_done` events, this contains the audio chunks that were processed.
+     * - For `end_of_speech` events, this contains the complete user speech.
+     */
+    frames: AudioFrame[];
+    /** Probability that speech is present (only for `INFERENCE_DONE` events). */
+    probability: number;
+    /** Time taken to perform the inference, in seconds (only for `INFERENCE_DONE` events). */
+    inferenceDuration: number;
+    /** Indicates whether speech was detected in the frames. */
+    speaking: boolean;
+    /** Threshold used to detect silence. */
+    rawAccumulatedSilence: number;
+    /** Threshold used to detect speech. */
+    rawAccumulatedSpeech: number;
+}
+export interface VADCapabilities {
+    updateInterval: number;
+}
+export type VADCallbacks = {
+    ['metrics_collected']: (metrics: VADMetrics) => void;
+};
+declare const VAD_base: new () => TypedEmitter<VADCallbacks>;
+export declare abstract class VAD extends VAD_base {
+    #private;
+    abstract label: string;
+    constructor(capabilities: VADCapabilities);
+    get capabilities(): VADCapabilities;
+    /**
+     * Returns a {@link VADStream} that can be used to push audio frames and receive VAD events.
+     */
+    abstract stream(): VADStream;
+    close(): Promise<void>;
+}
+export declare abstract class VADStream implements AsyncIterableIterator<VADEvent> {
+    #private;
+    protected static readonly FLUSH_SENTINEL: unique symbol;
+    protected input: IdentityTransform<AudioFrame | typeof VADStream.FLUSH_SENTINEL>;
+    protected output: IdentityTransform<VADEvent>;
+    protected inputWriter: WritableStreamDefaultWriter<AudioFrame | typeof VADStream.FLUSH_SENTINEL>;
+    protected inputReader: ReadableStreamDefaultReader<AudioFrame | typeof VADStream.FLUSH_SENTINEL>;
+    protected outputWriter: WritableStreamDefaultWriter<VADEvent>;
+    protected outputReader: ReadableStreamDefaultReader<VADEvent>;
+    protected closed: boolean;
+    protected inputClosed: boolean;
+    private logger;
+    private deferredInputStream;
+    private metricsStream;
+    constructor(vad: VAD);
+    /**
+     * Reads from the deferred input stream and forwards chunks to the input writer.
+     *
+     * Note: we can't just do this.deferredInputStream.stream.pipeTo(this.input.writable)
+     * because the inputWriter locks the this.input.writable stream. All writes must go through
+     * the inputWriter.
+     */
+    private pumpDeferredStream;
+    protected monitorMetrics(): Promise<void>;
+    /**
+     * Safely send a VAD event to the output stream, handling writer release errors during shutdown.
+     * @returns true if the event was sent, false if the stream is closing
+     * @throws Error if an unexpected error occurs
+     */
+    protected sendVADEvent(event: VADEvent): boolean;
+    updateInputStream(audioStream: ReadableStream<AudioFrame>): void;
+    detachInputStream(): void;
+    /** @deprecated Use `updateInputStream` instead */
+    pushFrame(frame: AudioFrame): void;
+    flush(): void;
+    endInput(): void;
+    next(): Promise<IteratorResult<VADEvent>>;
+    close(): void;
+    [Symbol.asyncIterator](): VADStream;
+}
+export {};
+//# sourceMappingURL=vad.d.ts.map

package/dist/vad.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"vad.d.ts","sourceRoot":"","sources":["../src/vad.ts"],"names":[],"mappings":";AAGA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AACpD,OAAO,KAAK,EAAE,iBAAiB,IAAI,YAAY,EAAE,MAAM,wBAAwB,CAAC;AAEhF,OAAO,KAAK,EACV,cAAc,EACd,2BAA2B,EAC3B,2BAA2B,EAC5B,MAAM,iBAAiB,CAAC;AAEzB,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAEpD,OAAO,EAAE,iBAAiB,EAAE,MAAM,gCAAgC,CAAC;AAEnE,oBAAY,YAAY;IACtB,eAAe,IAAA;IACf,cAAc,IAAA;IACd,aAAa,IAAA;IACb,iBAAiB,IAAA;CAClB;AAED,MAAM,WAAW,QAAQ;IACvB,oFAAoF;IACpF,IAAI,EAAE,YAAY,CAAC;IACnB;;OAEG;IACH,YAAY,EAAE,MAAM,CAAC;IACrB,0CAA0C;IAC1C,SAAS,EAAE,MAAM,CAAC;IAClB,sCAAsC;IACtC,cAAc,EAAE,MAAM,CAAC;IACvB,uCAAuC;IACvC,eAAe,EAAE,MAAM,CAAC;IACxB;;;;;;;OAOG;IACH,MAAM,EAAE,UAAU,EAAE,CAAC;IACrB,6EAA6E;IAC7E,WAAW,EAAE,MAAM,CAAC;IACpB,0FAA0F;IAC1F,iBAAiB,EAAE,MAAM,CAAC;IAC1B,2DAA2D;IAC3D,QAAQ,EAAE,OAAO,CAAC;IAClB,wCAAwC;IACxC,qBAAqB,EAAE,MAAM,CAAC;IAC9B,uCAAuC;IACvC,oBAAoB,EAAE,MAAM,CAAC;CAC9B;AAED,MAAM,WAAW,eAAe;IAC9B,cAAc,EAAE,MAAM,CAAC;CACxB;AAED,MAAM,MAAM,YAAY,GAAG;IACzB,CAAC,mBAAmB,CAAC,EAAE,CAAC,OAAO,EAAE,UAAU,KAAK,IAAI,CAAC;CACtD,CAAC;kCAE2D,aAAa,YAAY,CAAC;AAAvF,8BAAsB,GAAI,SAAQ,QAAsD;;IAEtF,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;gBAEX,YAAY,EAAE,eAAe;IAKzC,IAAI,YAAY,IAAI,eAAe,CAElC;IAED;;OAEG;IACH,QAAQ,CAAC,MAAM,IAAI,SAAS;IAEtB,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;CAG7B;AAED,8BAAsB,SAAU,YAAW,qBAAqB,CAAC,QAAQ,CAAC;;IACxE,SAAS,CAAC,MAAM,CAAC,QAAQ,CAAC,cAAc,gBAA4B;IACpE,SAAS,CAAC,KAAK,kEAAyE;IACxF,SAAS,CAAC,MAAM,8BAAqC;IACrD,SAAS,CAAC,WAAW,EAAE,2BAA2B,CAAC,UAAU,GAAG,OAAO,SAAS,CAAC,cAAc,CAAC,CAAC;IACjG,SAAS,CAAC,WAAW,EAAE,2BAA2B,CAAC,UAAU,GAAG,OAAO,SAAS,CAAC,cAAc,CAAC,CAAC;IACjG,SAAS,CAAC,YAAY,EAAE,2BAA2B,CAAC,QAAQ,CAAC,CAAC;IAC9D,SAAS,CAAC,YAAY,EAAE,2BAA2B,CAAC,QAAQ,CAAC,CAAC;IAC9D,SAAS,CAAC,MAAM,UAAS;IACzB,SAAS,CAAC,WAAW,UAAS;IAI9B,OAAO,CAAC,MAAM,CAAS;IACvB,OAAO,CAAC,mBAAmB,CAAqC;IAEhE,OAAO,CAAC,aAAa,CAA2B;gBACpC,GAAG,EAAE,GAAG;IAgBpB;;;;;;OAMG;YACW,kBAAkB;cAgBhB,cAAc;IAuC9B;;;;OAIG;IACH,SAAS,CAAC,YAAY,CAAC,KAAK,EAAE,QAAQ,GAAG,OAAO;IAahD,iBAAiB,CAAC,WAAW,EAAE,cAAc,CAAC,UAAU,CAAC;IAIzD,iBAAiB;IAIjB,kDAAkD;IAClD,SAAS,CAAC,KAAK,EAAE,UAAU;IAW3B,KAAK;IAUL,QAAQ;IAWF,IAAI,IAAI,OAAO,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC;IAS/C,KAAK;IAOL,CAAC,MAAM,CAAC,aAAa,CAAC,IAAI,SAAS;CAGpC"}

package/dist/vad.js

@@ -0,0 +1,185 @@
+import { EventEmitter } from "node:events";
+import { log } from "./log.js";
+import { DeferredReadableStream } from "./stream/deferred_stream.js";
+import { IdentityTransform } from "./stream/identity_transform.js";
+var VADEventType = /* @__PURE__ */ ((VADEventType2) => {
+  VADEventType2[VADEventType2["START_OF_SPEECH"] = 0] = "START_OF_SPEECH";
+  VADEventType2[VADEventType2["INFERENCE_DONE"] = 1] = "INFERENCE_DONE";
+  VADEventType2[VADEventType2["END_OF_SPEECH"] = 2] = "END_OF_SPEECH";
+  VADEventType2[VADEventType2["METRICS_COLLECTED"] = 3] = "METRICS_COLLECTED";
+  return VADEventType2;
+})(VADEventType || {});
+class VAD extends EventEmitter {
+  #capabilities;
+  constructor(capabilities) {
+    super();
+    this.#capabilities = capabilities;
+  }
+  get capabilities() {
+    return this.#capabilities;
+  }
+  async close() {
+    return;
+  }
+}
+class VADStream {
+  static FLUSH_SENTINEL = Symbol("FLUSH_SENTINEL");
+  input = new IdentityTransform();
+  output = new IdentityTransform();
+  inputWriter;
+  inputReader;
+  outputWriter;
+  outputReader;
+  closed = false;
+  inputClosed = false;
+  #vad;
+  #lastActivityTime = BigInt(0);
+  logger = log();
+  deferredInputStream;
+  metricsStream;
+  constructor(vad) {
+    this.#vad = vad;
+    this.deferredInputStream = new DeferredReadableStream();
+    this.inputWriter = this.input.writable.getWriter();
+    this.inputReader = this.input.readable.getReader();
+    this.outputWriter = this.output.writable.getWriter();
+    const [outputStream, metricsStream] = this.output.readable.tee();
+    this.metricsStream = metricsStream;
+    this.outputReader = outputStream.getReader();
+    this.pumpDeferredStream();
+    this.monitorMetrics();
+  }
+  /**
+   * Reads from the deferred input stream and forwards chunks to the input writer.
+   *
+   * Note: we can't just do this.deferredInputStream.stream.pipeTo(this.input.writable)
+   * because the inputWriter locks the this.input.writable stream. All writes must go through
+   * the inputWriter.
+   */
+  async pumpDeferredStream() {
+    const reader = this.deferredInputStream.stream.getReader();
+    try {
+      while (true) {
+        const { done, value } = await reader.read();
+        if (done) break;
+        await this.inputWriter.write(value);
+      }
+    } catch (e) {
+      this.logger.error(`Error pumping deferred stream: ${e}`);
+      throw e;
+    } finally {
+      reader.releaseLock();
+    }
+  }
+  async monitorMetrics() {
+    let inferenceDurationTotalMs = 0;
+    let inferenceCount = 0;
+    const metricsReader = this.metricsStream.getReader();
+    while (true) {
+      const { done, value } = await metricsReader.read();
+      if (done) {
+        break;
+      }
+      switch (value.type) {
+        case 0 /* START_OF_SPEECH */:
+          inferenceCount++;
+          if (inferenceCount >= 1 / this.#vad.capabilities.updateInterval) {
+            this.#vad.emit("metrics_collected", {
+              type: "vad_metrics",
+              timestamp: Date.now(),
+              idleTimeMs: Math.trunc(
+                Number((process.hrtime.bigint() - this.#lastActivityTime) / BigInt(1e6))
+              ),
+              inferenceDurationTotalMs,
+              inferenceCount,
+              label: this.#vad.label
+            });
+            inferenceCount = 0;
+            inferenceDurationTotalMs = 0;
+          }
+          break;
+        case 1 /* INFERENCE_DONE */:
+          inferenceDurationTotalMs += Math.round(value.inferenceDuration);
+          this.#lastActivityTime = process.hrtime.bigint();
+          break;
+        case 2 /* END_OF_SPEECH */:
+          this.#lastActivityTime = process.hrtime.bigint();
+          break;
+      }
+    }
+  }
+  /**
+   * Safely send a VAD event to the output stream, handling writer release errors during shutdown.
+   * @returns true if the event was sent, false if the stream is closing
+   * @throws Error if an unexpected error occurs
+   */
+  sendVADEvent(event) {
+    if (this.closed) {
+      return false;
+    }
+    try {
+      this.outputWriter.write(event);
+      return true;
+    } catch (e) {
+      throw e;
+    }
+  }
+  updateInputStream(audioStream) {
+    this.deferredInputStream.setSource(audioStream);
+  }
+  detachInputStream() {
+    this.deferredInputStream.detachSource();
+  }
+  /** @deprecated Use `updateInputStream` instead */
+  pushFrame(frame) {
+    if (this.inputClosed) {
+      throw new Error("Input is closed");
+    }
+    if (this.closed) {
+      throw new Error("Stream is closed");
+    }
+    this.inputWriter.write(frame);
+  }
+  flush() {
+    if (this.inputClosed) {
+      throw new Error("Input is closed");
+    }
+    if (this.closed) {
+      throw new Error("Stream is closed");
+    }
+    this.inputWriter.write(VADStream.FLUSH_SENTINEL);
+  }
+  endInput() {
+    if (this.inputClosed) {
+      throw new Error("Input is closed");
+    }
+    if (this.closed) {
+      throw new Error("Stream is closed");
+    }
+    this.inputClosed = true;
+    this.input.writable.close();
+  }
+  async next() {
+    return this.outputReader.read().then(({ done, value }) => {
+      if (done) {
+        return { done: true, value: void 0 };
+      }
+      return { done: false, value };
+    });
+  }
+  close() {
+    this.outputWriter.releaseLock();
+    this.outputReader.cancel();
+    this.output.writable.close();
+    this.closed = true;
+  }
+  [Symbol.asyncIterator]() {
+    return this;
+  }
+}
+export {
+  VAD,
+  VADEventType,
+  VADStream
+};
+//# sourceMappingURL=vad.js.map

package/dist/vad.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/vad.ts"],"sourcesContent":["// SPDX-FileCopyrightText: 2024 LiveKit, Inc.\n//\n// SPDX-License-Identifier: Apache-2.0\nimport type { AudioFrame } from '@livekit/rtc-node';\nimport type { TypedEventEmitter as TypedEmitter } from '@livekit/typed-emitter';\nimport { EventEmitter } from 'node:events';\nimport type {\n  ReadableStream,\n  ReadableStreamDefaultReader,\n  WritableStreamDefaultWriter,\n} from 'node:stream/web';\nimport { log } from './log.js';\nimport type { VADMetrics } from './metrics/base.js';\nimport { DeferredReadableStream } from './stream/deferred_stream.js';\nimport { IdentityTransform } from './stream/identity_transform.js';\n\nexport enum VADEventType {\n  START_OF_SPEECH,\n  INFERENCE_DONE,\n  END_OF_SPEECH,\n  METRICS_COLLECTED,\n}\n\nexport interface VADEvent {\n  /** Type of the VAD event (e.g., start of speech, end of speech, inference done). */\n  type: VADEventType;\n  /**\n   * Index of the audio sample where the event occurred, relative to the inference sample rate.\n   */\n  samplesIndex: number;\n  /** Timestamp when the event was fired. */\n  timestamp: number;\n  /** Duration of the speech segment. */\n  speechDuration: number;\n  /** Duration of the silence segment. */\n  silenceDuration: number;\n  /**\n   * List of audio frames associated with the speech.\n   *\n   * @remarks\n   * - For `start_of_speech` events, this contains the audio chunks that triggered the detection.\n   * - For `inference_done` events, this contains the audio chunks that were processed.\n   * - For `end_of_speech` events, this contains the complete user speech.\n   */\n  frames: AudioFrame[];\n  /** Probability that speech is present (only for `INFERENCE_DONE` events). */\n  probability: number;\n  /** Time taken to perform the inference, in seconds (only for `INFERENCE_DONE` events). */\n  inferenceDuration: number;\n  /** Indicates whether speech was detected in the frames. */\n  speaking: boolean;\n  /** Threshold used to detect silence. */\n  rawAccumulatedSilence: number;\n  /** Threshold used to detect speech. */\n  rawAccumulatedSpeech: number;\n}\n\nexport interface VADCapabilities {\n  updateInterval: number;\n}\n\nexport type VADCallbacks = {\n  ['metrics_collected']: (metrics: VADMetrics) => void;\n};\n\nexport abstract class VAD extends (EventEmitter as new () => TypedEmitter<VADCallbacks>) {\n  #capabilities: VADCapabilities;\n  abstract label: string;\n\n  constructor(capabilities: VADCapabilities) {\n    super();\n    this.#capabilities = capabilities;\n  }\n\n  get capabilities(): VADCapabilities {\n    return this.#capabilities;\n  }\n\n  /**\n   * Returns a {@link VADStream} that can be used to push audio frames and receive VAD events.\n   */\n  abstract stream(): VADStream;\n\n  async close(): Promise<void> {\n    return;\n  }\n}\n\nexport abstract class VADStream implements AsyncIterableIterator<VADEvent> {\n  protected static readonly FLUSH_SENTINEL = Symbol('FLUSH_SENTINEL');\n  protected input = new IdentityTransform<AudioFrame | typeof VADStream.FLUSH_SENTINEL>();\n  protected output = new IdentityTransform<VADEvent>();\n  protected inputWriter: WritableStreamDefaultWriter<AudioFrame | typeof VADStream.FLUSH_SENTINEL>;\n  protected inputReader: ReadableStreamDefaultReader<AudioFrame | typeof VADStream.FLUSH_SENTINEL>;\n  protected outputWriter: WritableStreamDefaultWriter<VADEvent>;\n  protected outputReader: ReadableStreamDefaultReader<VADEvent>;\n  protected closed = false;\n  protected inputClosed = false;\n\n  #vad: VAD;\n  #lastActivityTime = BigInt(0);\n  private logger = log();\n  private deferredInputStream: DeferredReadableStream<AudioFrame>;\n\n  private metricsStream: ReadableStream<VADEvent>;\n  constructor(vad: VAD) {\n    this.#vad = vad;\n    this.deferredInputStream = new DeferredReadableStream<AudioFrame>();\n\n    this.inputWriter = this.input.writable.getWriter();\n    this.inputReader = this.input.readable.getReader();\n    this.outputWriter = this.output.writable.getWriter();\n\n    const [outputStream, metricsStream] = this.output.readable.tee();\n    this.metricsStream = metricsStream;\n    this.outputReader = outputStream.getReader();\n\n    this.pumpDeferredStream();\n    this.monitorMetrics();\n  }\n\n  /**\n   * Reads from the deferred input stream and forwards chunks to the input writer.\n   *\n   * Note: we can't just do this.deferredInputStream.stream.pipeTo(this.input.writable)\n   * because the inputWriter locks the this.input.writable stream. All writes must go through\n   * the inputWriter.\n   */\n  private async pumpDeferredStream() {\n    const reader = this.deferredInputStream.stream.getReader();\n    try {\n      while (true) {\n        const { done, value } = await reader.read();\n        if (done) break;\n        await this.inputWriter.write(value);\n      }\n    } catch (e) {\n      this.logger.error(`Error pumping deferred stream: ${e}`);\n      throw e;\n    } finally {\n      reader.releaseLock();\n    }\n  }\n\n  protected async monitorMetrics() {\n    let inferenceDurationTotalMs = 0;\n    let inferenceCount = 0;\n    const metricsReader = this.metricsStream.getReader();\n    while (true) {\n      const { done, value } = await metricsReader.read();\n      if (done) {\n        break;\n      }\n      switch (value.type) {\n        case VADEventType.START_OF_SPEECH:\n          inferenceCount++;\n          if (inferenceCount >= 1 / this.#vad.capabilities.updateInterval) {\n            this.#vad.emit('metrics_collected', {\n              type: 'vad_metrics',\n              timestamp: Date.now(),\n              idleTimeMs: Math.trunc(\n                Number((process.hrtime.bigint() - this.#lastActivityTime) / BigInt(1000000)),\n              ),\n              inferenceDurationTotalMs,\n              inferenceCount,\n              label: this.#vad.label,\n            });\n\n            inferenceCount = 0;\n            inferenceDurationTotalMs = 0;\n          }\n          break;\n        case VADEventType.INFERENCE_DONE:\n          inferenceDurationTotalMs += Math.round(value.inferenceDuration);\n          this.#lastActivityTime = process.hrtime.bigint();\n          break;\n        case VADEventType.END_OF_SPEECH:\n          this.#lastActivityTime = process.hrtime.bigint();\n          break;\n      }\n    }\n  }\n\n  /**\n   * Safely send a VAD event to the output stream, handling writer release errors during shutdown.\n   * @returns true if the event was sent, false if the stream is closing\n   * @throws Error if an unexpected error occurs\n   */\n  protected sendVADEvent(event: VADEvent): boolean {\n    if (this.closed) {\n      return false;\n    }\n\n    try {\n      this.outputWriter.write(event);\n      return true;\n    } catch (e) {\n      throw e;\n    }\n  }\n\n  updateInputStream(audioStream: ReadableStream<AudioFrame>) {\n    this.deferredInputStream.setSource(audioStream);\n  }\n\n  detachInputStream() {\n    this.deferredInputStream.detachSource();\n  }\n\n  /** @deprecated Use `updateInputStream` instead */\n  pushFrame(frame: AudioFrame) {\n    // TODO(AJS-395): remove this method\n    if (this.inputClosed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.inputWriter.write(frame);\n  }\n\n  flush() {\n    if (this.inputClosed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.inputWriter.write(VADStream.FLUSH_SENTINEL);\n  }\n\n  endInput() {\n    if (this.inputClosed) {\n      throw new Error('Input is closed');\n    }\n    if (this.closed) {\n      throw new Error('Stream is closed');\n    }\n    this.inputClosed = true;\n    this.input.writable.close();\n  }\n\n  async next(): Promise<IteratorResult<VADEvent>> {\n    return this.outputReader.read().then(({ done, value }) => {\n      if (done) {\n        return { done: true, value: undefined };\n      }\n      return { done: false, value };\n    });\n  }\n\n  close() {\n    this.outputWriter.releaseLock();\n    this.outputReader.cancel();\n    this.output.writable.close();\n    this.closed = true;\n  }\n\n  [Symbol.asyncIterator](): VADStream {\n    return this;\n  }\n}\n"],"mappings":"AAKA,SAAS,oBAAoB;AAM7B,SAAS,WAAW;AAEpB,SAAS,8BAA8B;AACvC,SAAS,yBAAyB;AAE3B,IAAK,eAAL,kBAAKA,kBAAL;AACL,EAAAA,4BAAA;AACA,EAAAA,4BAAA;AACA,EAAAA,4BAAA;AACA,EAAAA,4BAAA;AAJU,SAAAA;AAAA,GAAA;AAiDL,MAAe,YAAa,aAAsD;AAAA,EACvF;AAAA,EAGA,YAAY,cAA+B;AACzC,UAAM;AACN,SAAK,gBAAgB;AAAA,EACvB;AAAA,EAEA,IAAI,eAAgC;AAClC,WAAO,KAAK;AAAA,EACd;AAAA,EAOA,MAAM,QAAuB;AAC3B;AAAA,EACF;AACF;AAEO,MAAe,UAAqD;AAAA,EACzE,OAA0B,iBAAiB,OAAO,gBAAgB;AAAA,EACxD,QAAQ,IAAI,kBAAgE;AAAA,EAC5E,SAAS,IAAI,kBAA4B;AAAA,EACzC;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA,SAAS;AAAA,EACT,cAAc;AAAA,EAExB;AAAA,EACA,oBAAoB,OAAO,CAAC;AAAA,EACpB,SAAS,IAAI;AAAA,EACb;AAAA,EAEA;AAAA,EACR,YAAY,KAAU;AACpB,SAAK,OAAO;AACZ,SAAK,sBAAsB,IAAI,uBAAmC;AAElE,SAAK,cAAc,KAAK,MAAM,SAAS,UAAU;AACjD,SAAK,cAAc,KAAK,MAAM,SAAS,UAAU;AACjD,SAAK,eAAe,KAAK,OAAO,SAAS,UAAU;AAEnD,UAAM,CAAC,cAAc,aAAa,IAAI,KAAK,OAAO,SAAS,IAAI;AAC/D,SAAK,gBAAgB;AACrB,SAAK,eAAe,aAAa,UAAU;AAE3C,SAAK,mBAAmB;AACxB,SAAK,eAAe;AAAA,EACtB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EASA,MAAc,qBAAqB;AACjC,UAAM,SAAS,KAAK,oBAAoB,OAAO,UAAU;AACzD,QAAI;AACF,aAAO,MAAM;AACX,cAAM,EAAE,MAAM,MAAM,IAAI,MAAM,OAAO,KAAK;AAC1C,YAAI,KAAM;AACV,cAAM,KAAK,YAAY,MAAM,KAAK;AAAA,MACpC;AAAA,IACF,SAAS,GAAG;AACV,WAAK,OAAO,MAAM,kCAAkC,CAAC,EAAE;AACvD,YAAM;AAAA,IACR,UAAE;AACA,aAAO,YAAY;AAAA,IACrB;AAAA,EACF;AAAA,EAEA,MAAgB,iBAAiB;AAC/B,QAAI,2BAA2B;AAC/B,QAAI,iBAAiB;AACrB,UAAM,gBAAgB,KAAK,cAAc,UAAU;AACnD,WAAO,MAAM;AACX,YAAM,EAAE,MAAM,MAAM,IAAI,MAAM,cAAc,KAAK;AACjD,UAAI,MAAM;AACR;AAAA,MACF;AACA,cAAQ,MAAM,MAAM;AAAA,QAClB,KAAK;AACH;AACA,cAAI,kBAAkB,IAAI,KAAK,KAAK,aAAa,gBAAgB;AAC/D,iBAAK,KAAK,KAAK,qBAAqB;AAAA,cAClC,MAAM;AAAA,cACN,WAAW,KAAK,IAAI;AAAA,cACpB,YAAY,KAAK;AAAA,gBACf,QAAQ,QAAQ,OAAO,OAAO,IAAI,KAAK,qBAAqB,OAAO,GAAO,CAAC;AAAA,cAC7E;AAAA,cACA;AAAA,cACA;AAAA,cACA,OAAO,KAAK,KAAK;AAAA,YACnB,CAAC;AAED,6BAAiB;AACjB,uCAA2B;AAAA,UAC7B;AACA;AAAA,QACF,KAAK;AACH,sCAA4B,KAAK,MAAM,MAAM,iBAAiB;AAC9D,eAAK,oBAAoB,QAAQ,OAAO,OAAO;AAC/C;AAAA,QACF,KAAK;AACH,eAAK,oBAAoB,QAAQ,OAAO,OAAO;AAC/C;AAAA,MACJ;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOU,aAAa,OAA0B;AAC/C,QAAI,KAAK,QAAQ;AACf,aAAO;AAAA,IACT;AAEA,QAAI;AACF,WAAK,aAAa,MAAM,KAAK;AAC7B,aAAO;AAAA,IACT,SAAS,GAAG;AACV,YAAM;AAAA,IACR;AAAA,EACF;AAAA,EAEA,kBAAkB,aAAyC;AACzD,SAAK,oBAAoB,UAAU,WAAW;AAAA,EAChD;AAAA,EAEA,oBAAoB;AAClB,SAAK,oBAAoB,aAAa;AAAA,EACxC;AAAA;AAAA,EAGA,UAAU,OAAmB;AAE3B,QAAI,KAAK,aAAa;AACpB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,YAAY,MAAM,KAAK;AAAA,EAC9B;AAAA,EAEA,QAAQ;AACN,QAAI,KAAK,aAAa;AACpB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,YAAY,MAAM,UAAU,cAAc;AAAA,EACjD;AAAA,EAEA,WAAW;AACT,QAAI,KAAK,aAAa;AACpB,YAAM,IAAI,MAAM,iBAAiB;AAAA,IACnC;AACA,QAAI,KAAK,QAAQ;AACf,YAAM,IAAI,MAAM,kBAAkB;AAAA,IACpC;AACA,SAAK,cAAc;AACnB,SAAK,MAAM,SAAS,MAAM;AAAA,EAC5B;AAAA,EAEA,MAAM,OAA0C;AAC9C,WAAO,KAAK,aAAa,KAAK,EAAE,KAAK,CAAC,EAAE,MAAM,MAAM,MAAM;AACxD,UAAI,MAAM;AACR,eAAO,EAAE,MAAM,MAAM,OAAO,OAAU;AAAA,MACxC;AACA,aAAO,EAAE,MAAM,OAAO,MAAM;AAAA,IAC9B,CAAC;AAAA,EACH;AAAA,EAEA,QAAQ;AACN,SAAK,aAAa,YAAY;AAC9B,SAAK,aAAa,OAAO;AACzB,SAAK,OAAO,SAAS,MAAM;AAC3B,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,CAAC,OAAO,aAAa,IAAe;AAClC,WAAO;AAAA,EACT;AACF;","names":["VADEventType"]}