@livekit/agents 0.0.0-20260120144724

package/dist/voice/audio_recognition.cjs

@@ -0,0 +1,668 @@
+"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 audio_recognition_exports = {};
+__export(audio_recognition_exports, {
+  AudioRecognition: () => AudioRecognition
+});
+module.exports = __toCommonJS(audio_recognition_exports);
+var import_rtc_node = require("@livekit/rtc-node");
+var import_web = require("node:stream/web");
+var import_InterruptionStream = require("../inference/interruption/InterruptionStream.cjs");
+var import_chat_context = require("../llm/chat_context.cjs");
+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 import_merge_readable_streams = require("../stream/merge_readable_streams.cjs");
+var import_stt = require("../stt/stt.cjs");
+var import_telemetry = require("../telemetry/index.cjs");
+var import_utils = require("../utils.cjs");
+var import_vad = require("../vad.cjs");
+class AudioRecognition {
+  hooks;
+  stt;
+  vad;
+  turnDetector;
+  turnDetectionMode;
+  minEndpointingDelay;
+  maxEndpointingDelay;
+  lastLanguage;
+  rootSpanContext;
+  deferredInputStream;
+  logger = (0, import_log.log)();
+  lastFinalTranscriptTime = 0;
+  audioTranscript = "";
+  audioInterimTranscript = "";
+  audioPreflightTranscript = "";
+  finalTranscriptConfidence = [];
+  lastSpeakingTime;
+  speechStartTime;
+  userTurnCommitted = false;
+  speaking = false;
+  sampleRate;
+  userTurnSpan;
+  vadInputStream;
+  sttInputStream;
+  interruptionInputStream;
+  silenceAudioTransform = new import_identity_transform.IdentityTransform();
+  silenceAudioWriter;
+  // all cancellable tasks
+  bounceEOUTask;
+  commitUserTurnTask;
+  vadTask;
+  sttTask;
+  interruptionTask;
+  // interruption detection
+  interruptionDetector;
+  interruptionStream;
+  interruptionEnabled = false;
+  agentSpeaking = false;
+  constructor(opts) {
+    this.hooks = opts.recognitionHooks;
+    this.stt = opts.stt;
+    this.vad = opts.vad;
+    this.interruptionDetector = opts.interruptionDetector;
+    this.turnDetector = opts.turnDetector;
+    this.turnDetectionMode = opts.turnDetectionMode;
+    this.minEndpointingDelay = opts.minEndpointingDelay;
+    this.maxEndpointingDelay = opts.maxEndpointingDelay;
+    this.lastLanguage = void 0;
+    this.rootSpanContext = opts.rootSpanContext;
+    this.interruptionEnabled = this.interruptionDetector !== void 0 && this.vad !== void 0;
+    this.deferredInputStream = new import_deferred_stream.DeferredReadableStream();
+    const [vadInputStream, rest] = this.deferredInputStream.stream.tee();
+    const [sttInputStream, interruptionInputStream] = rest.tee();
+    this.vadInputStream = vadInputStream;
+    this.sttInputStream = (0, import_merge_readable_streams.mergeReadableStreams)(sttInputStream, this.silenceAudioTransform.readable);
+    this.interruptionInputStream = interruptionInputStream;
+    this.silenceAudioWriter = this.silenceAudioTransform.writable.getWriter();
+  }
+  /**
+   * Current transcript of the user's speech, including interim transcript if available.
+   */
+  get currentTranscript() {
+    if (this.audioInterimTranscript) {
+      return `${this.audioTranscript} ${this.audioInterimTranscript}`.trim();
+    }
+    return this.audioTranscript;
+  }
+  async start() {
+    this.vadTask = import_utils.Task.from(({ signal }) => this.createVadTask(this.vad, signal));
+    this.vadTask.result.catch((err) => {
+      this.logger.error(`Error running VAD task: ${err}`);
+    });
+    this.sttTask = import_utils.Task.from(({ signal }) => this.createSttTask(this.stt, signal));
+    this.sttTask.result.catch((err) => {
+      this.logger.error(`Error running STT task: ${err}`);
+    });
+    if (this.interruptionEnabled && this.interruptionDetector) {
+      this.interruptionTask = import_utils.Task.from(
+        ({ signal }) => this.createInterruptionTask(this.interruptionDetector, signal)
+      );
+      this.interruptionTask.result.catch((err) => {
+        this.logger.error(`Error running interruption task: ${err}`);
+      });
+    }
+  }
+  async onSTTEvent(ev) {
+    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k, _l, _m, _n, _o, _p, _q, _r;
+    if (this.turnDetectionMode === "manual" && this.userTurnCommitted && (this.bounceEOUTask === void 0 || this.bounceEOUTask.done || ev.type == import_stt.SpeechEventType.INTERIM_TRANSCRIPT)) {
+      this.logger.debug(
+        {
+          userTurnCommitted: this.userTurnCommitted,
+          eouTaskDone: (_a = this.bounceEOUTask) == null ? void 0 : _a.done,
+          evType: ev.type,
+          turnDetectionMode: this.turnDetectionMode
+        },
+        "ignoring stt event"
+      );
+      return;
+    }
+    switch (ev.type) {
+      case import_stt.SpeechEventType.FINAL_TRANSCRIPT:
+        this.hooks.onFinalTranscript(ev);
+        const transcript = (_c = (_b = ev.alternatives) == null ? void 0 : _b[0]) == null ? void 0 : _c.text;
+        const confidence = ((_e = (_d = ev.alternatives) == null ? void 0 : _d[0]) == null ? void 0 : _e.confidence) ?? 0;
+        this.lastLanguage = (_g = (_f = ev.alternatives) == null ? void 0 : _f[0]) == null ? void 0 : _g.language;
+        if (!transcript) {
+          return;
+        }
+        this.logger.debug(
+          {
+            user_transcript: transcript,
+            language: this.lastLanguage
+          },
+          "received user transcript"
+        );
+        this.lastFinalTranscriptTime = Date.now();
+        this.audioTranscript += ` ${transcript}`;
+        this.audioTranscript = this.audioTranscript.trimStart();
+        this.finalTranscriptConfidence.push(confidence);
+        const transcriptChanged = this.audioTranscript !== this.audioPreflightTranscript;
+        this.audioInterimTranscript = "";
+        this.audioPreflightTranscript = "";
+        if (!this.vad || this.lastSpeakingTime === void 0) {
+          this.lastSpeakingTime = Date.now();
+        }
+        if (this.vadBaseTurnDetection || this.userTurnCommitted) {
+          if (transcriptChanged) {
+            this.logger.debug(
+              { transcript: this.audioTranscript },
+              "triggering preemptive generation (FINAL_TRANSCRIPT)"
+            );
+            this.hooks.onPreemptiveGeneration({
+              newTranscript: this.audioTranscript,
+              transcriptConfidence: this.finalTranscriptConfidence.length > 0 ? this.finalTranscriptConfidence.reduce((a, b) => a + b, 0) / this.finalTranscriptConfidence.length : 0
+            });
+          }
+          if (!this.speaking) {
+            const chatCtx = this.hooks.retrieveChatCtx();
+            this.logger.debug("running EOU detection on stt FINAL_TRANSCRIPT");
+            this.runEOUDetection(chatCtx);
+          }
+        }
+        break;
+      case import_stt.SpeechEventType.PREFLIGHT_TRANSCRIPT:
+        this.hooks.onInterimTranscript(ev);
+        const preflightTranscript = ((_i = (_h = ev.alternatives) == null ? void 0 : _h[0]) == null ? void 0 : _i.text) ?? "";
+        const preflightConfidence = ((_k = (_j = ev.alternatives) == null ? void 0 : _j[0]) == null ? void 0 : _k.confidence) ?? 0;
+        const preflightLanguage = (_m = (_l = ev.alternatives) == null ? void 0 : _l[0]) == null ? void 0 : _m.language;
+        const MIN_LANGUAGE_DETECTION_LENGTH = 5;
+        if (!this.lastLanguage || preflightLanguage && preflightTranscript.length > MIN_LANGUAGE_DETECTION_LENGTH) {
+          this.lastLanguage = preflightLanguage;
+        }
+        if (!preflightTranscript) {
+          return;
+        }
+        this.logger.debug(
+          {
+            user_transcript: preflightTranscript,
+            language: this.lastLanguage
+          },
+          "received user preflight transcript"
+        );
+        this.lastFinalTranscriptTime = Date.now();
+        this.audioPreflightTranscript = `${this.audioTranscript} ${preflightTranscript}`.trimStart();
+        this.audioInterimTranscript = preflightTranscript;
+        if (!this.vad || this.lastSpeakingTime === void 0) {
+          this.lastSpeakingTime = Date.now();
+        }
+        if (this.turnDetectionMode !== "manual" || this.userTurnCommitted) {
+          const confidenceVals = [...this.finalTranscriptConfidence, preflightConfidence];
+          this.logger.debug(
+            {
+              transcript: this.audioPreflightTranscript.length > 100 ? this.audioPreflightTranscript.slice(0, 100) + "..." : this.audioPreflightTranscript
+            },
+            "triggering preemptive generation (PREFLIGHT_TRANSCRIPT)"
+          );
+          this.hooks.onPreemptiveGeneration({
+            newTranscript: this.audioPreflightTranscript,
+            transcriptConfidence: confidenceVals.length > 0 ? confidenceVals.reduce((a, b) => a + b, 0) / confidenceVals.length : 0
+          });
+        }
+        break;
+      case import_stt.SpeechEventType.INTERIM_TRANSCRIPT:
+        this.logger.debug({ transcript: (_o = (_n = ev.alternatives) == null ? void 0 : _n[0]) == null ? void 0 : _o.text }, "interim transcript");
+        this.hooks.onInterimTranscript(ev);
+        this.audioInterimTranscript = ((_q = (_p = ev.alternatives) == null ? void 0 : _p[0]) == null ? void 0 : _q.text) ?? "";
+        break;
+      case import_stt.SpeechEventType.START_OF_SPEECH:
+        if (this.turnDetectionMode !== "stt") break;
+        this.hooks.onStartOfSpeech({
+          type: import_vad.VADEventType.START_OF_SPEECH,
+          samplesIndex: 0,
+          timestamp: Date.now(),
+          speechDuration: 0,
+          silenceDuration: 0,
+          frames: [],
+          probability: 0,
+          inferenceDuration: 0,
+          speaking: true,
+          rawAccumulatedSilence: 0,
+          rawAccumulatedSpeech: 0
+        });
+        this.speaking = true;
+        this.lastSpeakingTime = Date.now();
+        (_r = this.bounceEOUTask) == null ? void 0 : _r.cancel();
+        break;
+      case import_stt.SpeechEventType.END_OF_SPEECH:
+        if (this.turnDetectionMode !== "stt") break;
+        this.hooks.onEndOfSpeech({
+          type: import_vad.VADEventType.END_OF_SPEECH,
+          samplesIndex: 0,
+          timestamp: Date.now(),
+          speechDuration: 0,
+          silenceDuration: 0,
+          frames: [],
+          probability: 0,
+          inferenceDuration: 0,
+          speaking: false,
+          rawAccumulatedSilence: 0,
+          rawAccumulatedSpeech: 0
+        });
+        this.speaking = false;
+        this.userTurnCommitted = true;
+        this.lastSpeakingTime = Date.now();
+        if (!this.speaking) {
+          const chatCtx = this.hooks.retrieveChatCtx();
+          this.logger.debug("running EOU detection on stt END_OF_SPEECH");
+          this.runEOUDetection(chatCtx);
+        }
+    }
+  }
+  runEOUDetection(chatCtx) {
+    var _a;
+    this.logger.debug(
+      {
+        stt: this.stt,
+        audioTranscript: this.audioTranscript,
+        turnDetectionMode: this.turnDetectionMode
+      },
+      "running EOU detection"
+    );
+    if (this.stt && !this.audioTranscript && this.turnDetectionMode !== "manual") {
+      this.logger.debug("skipping EOU detection");
+      return;
+    }
+    chatCtx = chatCtx.copy();
+    chatCtx.addMessage({ role: "user", content: this.audioTranscript });
+    const turnDetector = (
+      // disable EOU model if manual turn detection enabled
+      this.audioTranscript && this.turnDetectionMode !== "manual" ? this.turnDetector : void 0
+    );
+    const bounceEOUTask = (lastSpeakingTime, lastFinalTranscriptTime, speechStartTime) => async (controller) => {
+      let endpointingDelay = this.minEndpointingDelay;
+      if (turnDetector) {
+        await import_telemetry.tracer.startActiveSpan(
+          async (span) => {
+            this.logger.debug("Running turn detector model");
+            let endOfTurnProbability = 0;
+            let unlikelyThreshold;
+            if (!await turnDetector.supportsLanguage(this.lastLanguage)) {
+              this.logger.debug(`Turn detector does not support language ${this.lastLanguage}`);
+            } else {
+              try {
+                endOfTurnProbability = await turnDetector.predictEndOfTurn(chatCtx);
+                unlikelyThreshold = await turnDetector.unlikelyThreshold(this.lastLanguage);
+                this.logger.debug(
+                  { endOfTurnProbability, unlikelyThreshold, language: this.lastLanguage },
+                  "end of turn probability"
+                );
+                if (unlikelyThreshold && endOfTurnProbability < unlikelyThreshold) {
+                  endpointingDelay = this.maxEndpointingDelay;
+                }
+              } catch (error) {
+                this.logger.error(error, "Error predicting end of turn");
+              }
+            }
+            span.setAttribute(
+              import_telemetry.traceTypes.ATTR_CHAT_CTX,
+              JSON.stringify(chatCtx.toJSON({ excludeTimestamp: false }))
+            );
+            span.setAttribute(import_telemetry.traceTypes.ATTR_EOU_PROBABILITY, endOfTurnProbability);
+            span.setAttribute(import_telemetry.traceTypes.ATTR_EOU_UNLIKELY_THRESHOLD, unlikelyThreshold ?? 0);
+            span.setAttribute(import_telemetry.traceTypes.ATTR_EOU_DELAY, endpointingDelay);
+            span.setAttribute(import_telemetry.traceTypes.ATTR_EOU_LANGUAGE, this.lastLanguage ?? "");
+          },
+          {
+            name: "eou_detection",
+            context: this.rootSpanContext
+          }
+        );
+      }
+      let extraSleep = endpointingDelay;
+      if (lastSpeakingTime !== void 0) {
+        extraSleep += lastSpeakingTime - Date.now();
+      }
+      if (extraSleep > 0) {
+        await (0, import_utils.delay)(Math.max(extraSleep, 0), { signal: controller.signal });
+      }
+      this.logger.debug({ transcript: this.audioTranscript }, "end of user turn");
+      const confidenceAvg = this.finalTranscriptConfidence.length > 0 ? this.finalTranscriptConfidence.reduce((a, b) => a + b, 0) / this.finalTranscriptConfidence.length : 0;
+      let startedSpeakingAt;
+      let stoppedSpeakingAt;
+      let transcriptionDelay;
+      let endOfUtteranceDelay;
+      if (lastFinalTranscriptTime !== 0 && lastSpeakingTime !== void 0 && speechStartTime !== void 0) {
+        startedSpeakingAt = speechStartTime;
+        stoppedSpeakingAt = lastSpeakingTime;
+        transcriptionDelay = Math.max(lastFinalTranscriptTime - lastSpeakingTime, 0);
+        endOfUtteranceDelay = Date.now() - lastSpeakingTime;
+      }
+      const committed = await this.hooks.onEndOfTurn({
+        newTranscript: this.audioTranscript,
+        transcriptConfidence: confidenceAvg,
+        transcriptionDelay: transcriptionDelay ?? 0,
+        endOfUtteranceDelay: endOfUtteranceDelay ?? 0,
+        startedSpeakingAt,
+        stoppedSpeakingAt
+      });
+      if (committed) {
+        this._endUserTurnSpan({
+          transcript: this.audioTranscript,
+          confidence: confidenceAvg,
+          transcriptionDelay: transcriptionDelay ?? 0,
+          endOfUtteranceDelay: endOfUtteranceDelay ?? 0
+        });
+        this.audioTranscript = "";
+        this.finalTranscriptConfidence = [];
+        this.lastSpeakingTime = void 0;
+        this.lastFinalTranscriptTime = 0;
+        this.speechStartTime = void 0;
+      }
+      this.userTurnCommitted = false;
+    };
+    (_a = this.bounceEOUTask) == null ? void 0 : _a.cancel();
+    this.bounceEOUTask = import_utils.Task.from(
+      bounceEOUTask(this.lastSpeakingTime, this.lastFinalTranscriptTime, this.speechStartTime)
+    );
+    this.bounceEOUTask.result.then(() => {
+      this.logger.debug("EOU detection task completed");
+    }).catch((err) => {
+      if (err instanceof Error && err.message.includes("This operation was aborted")) {
+        return;
+      }
+      this.logger.error(err, "Error in EOU detection task:");
+    });
+  }
+  async createSttTask(stt, signal) {
+    if (!stt) return;
+    this.logger.debug("createSttTask: create stt stream from stt node");
+    const sttStream = await stt(this.sttInputStream, {});
+    if (signal.aborted || sttStream === null) return;
+    if (sttStream instanceof import_web.ReadableStream) {
+      const reader = sttStream.getReader();
+      signal.addEventListener("abort", async () => {
+        try {
+          reader.releaseLock();
+          await (sttStream == null ? void 0 : sttStream.cancel());
+        } catch (e) {
+          this.logger.debug("createSttTask: error during abort handler:", e);
+        }
+      });
+      try {
+        while (true) {
+          if (signal.aborted) break;
+          const { done, value: ev } = await reader.read();
+          if (done) break;
+          if (typeof ev === "string") {
+            throw new Error("STT node must yield SpeechEvent");
+          } else {
+            await this.onSTTEvent(ev);
+          }
+        }
+      } catch (e) {
+        if ((0, import_deferred_stream.isStreamReaderReleaseError)(e)) {
+          return;
+        }
+        this.logger.error({ error: e }, "createSttTask: error reading sttStream");
+      } finally {
+        reader.releaseLock();
+        try {
+          await sttStream.cancel();
+        } catch (e) {
+          this.logger.debug(
+            "createSttTask: error cancelling sttStream (may already be cancelled):",
+            e
+          );
+        }
+      }
+    }
+  }
+  async createVadTask(vad, signal) {
+    var _a;
+    if (!vad) return;
+    const vadStream = vad.stream();
+    vadStream.updateInputStream(this.vadInputStream);
+    const abortHandler = () => {
+      vadStream.detachInputStream();
+      vadStream.close();
+      signal.removeEventListener("abort", abortHandler);
+    };
+    signal.addEventListener("abort", abortHandler);
+    try {
+      for await (const ev of vadStream) {
+        if (signal.aborted) break;
+        switch (ev.type) {
+          case import_vad.VADEventType.START_OF_SPEECH:
+            this.logger.debug("VAD task: START_OF_SPEECH");
+            this.hooks.onStartOfSpeech(ev);
+            this.speaking = true;
+            if (!this.userTurnSpan) {
+              this.userTurnSpan = import_telemetry.tracer.startSpan({
+                name: "user_turn",
+                context: this.rootSpanContext
+              });
+            }
+            if (ev.frames.length > 0 && ev.frames[0]) {
+              this.sampleRate = ev.frames[0].sampleRate;
+            }
+            if (this.agentSpeaking && this.interruptionEnabled) {
+              this.onStartOfOverlapSpeech(ev.speechDuration, this.userTurnSpan);
+            }
+            (_a = this.bounceEOUTask) == null ? void 0 : _a.cancel();
+            break;
+          case import_vad.VADEventType.INFERENCE_DONE:
+            this.hooks.onVADInferenceDone(ev);
+            if (ev.rawAccumulatedSpeech > 0) {
+              this.lastSpeakingTime = Date.now();
+              if (this.speechStartTime === void 0) {
+                this.speechStartTime = Date.now();
+              }
+            }
+            break;
+          case import_vad.VADEventType.END_OF_SPEECH:
+            this.logger.debug("VAD task: END_OF_SPEECH");
+            this.hooks.onEndOfSpeech(ev);
+            this.speaking = false;
+            if (this.agentSpeaking && this.interruptionEnabled) {
+              this.onEndOfOverlapSpeech();
+            }
+            if (this.vadBaseTurnDetection || this.turnDetectionMode === "stt" && this.userTurnCommitted) {
+              const chatCtx = this.hooks.retrieveChatCtx();
+              this.runEOUDetection(chatCtx);
+            }
+            break;
+        }
+      }
+    } catch (e) {
+      this.logger.error(e, "Error in VAD task");
+    } finally {
+      this.logger.debug("VAD task closed");
+    }
+  }
+  async createInterruptionTask(interruptionDetector, signal) {
+    this.interruptionStream = interruptionDetector.createStream();
+    const reader = this.interruptionInputStream.getReader();
+    const forwardTask = (async () => {
+      var _a;
+      try {
+        while (!signal.aborted) {
+          const { done, value: frame } = await reader.read();
+          if (done) break;
+          await ((_a = this.interruptionStream) == null ? void 0 : _a.pushFrame(frame));
+        }
+      } catch (e) {
+        if (!signal.aborted) {
+          this.logger.error(e, "Error forwarding audio to interruption stream");
+        }
+      } finally {
+        reader.releaseLock();
+      }
+    })();
+    const eventStream = this.interruptionStream.stream;
+    const eventReader = eventStream.getReader();
+    const abortHandler = () => {
+      var _a;
+      eventReader.releaseLock();
+      (_a = this.interruptionStream) == null ? void 0 : _a.close();
+      signal.removeEventListener("abort", abortHandler);
+    };
+    signal.addEventListener("abort", abortHandler);
+    try {
+      while (!signal.aborted) {
+        const { done, value: ev } = await eventReader.read();
+        if (done) break;
+        this.logger.debug({ type: ev.type, probability: ev.probability }, "Interruption event");
+        this.hooks.onInterruption(ev);
+      }
+    } catch (e) {
+      if (!signal.aborted) {
+        this.logger.error(e, "Error in interruption task");
+      }
+    } finally {
+      this.logger.debug("Interruption task closed");
+      await forwardTask;
+    }
+  }
+  /**
+   * Called when the agent starts speaking.
+   * Enables interruption detection by sending the agent-speech-started sentinel.
+   */
+  onStartOfAgentSpeech() {
+    this.agentSpeaking = true;
+    if (!this.interruptionEnabled || !this.interruptionStream) {
+      return;
+    }
+    this.interruptionStream.pushFrame(import_InterruptionStream.InterruptionStreamSentinel.speechStarted());
+  }
+  /**
+   * Called when the agent stops speaking.
+   * Disables interruption detection by sending the agent-speech-ended sentinel.
+   */
+  onEndOfAgentSpeech() {
+    if (!this.interruptionEnabled || !this.interruptionStream) {
+      this.agentSpeaking = false;
+      return;
+    }
+    this.interruptionStream.pushFrame(import_InterruptionStream.InterruptionStreamSentinel.speechEnded());
+    if (this.agentSpeaking) {
+      this.onEndOfOverlapSpeech();
+    }
+    this.agentSpeaking = false;
+  }
+  /**
+   * Called when user starts speaking while agent is speaking (overlap speech).
+   * This triggers the interruption detection inference.
+   */
+  onStartOfOverlapSpeech(speechDuration, userSpeakingSpan) {
+    if (!this.interruptionEnabled || !this.interruptionStream) {
+      return;
+    }
+    if (this.agentSpeaking && userSpeakingSpan) {
+      this.interruptionStream.pushFrame(
+        import_InterruptionStream.InterruptionStreamSentinel.overlapSpeechStarted(speechDuration, userSpeakingSpan)
+      );
+    }
+  }
+  /**
+   * Called when user stops speaking during overlap.
+   * This ends the interruption detection inference for this overlap period.
+   */
+  onEndOfOverlapSpeech() {
+    if (!this.interruptionEnabled || !this.interruptionStream) {
+      return;
+    }
+    this.interruptionStream.pushFrame(import_InterruptionStream.InterruptionStreamSentinel.overlapSpeechEnded());
+  }
+  setInputAudioStream(audioStream) {
+    this.deferredInputStream.setSource(audioStream);
+  }
+  detachInputAudioStream() {
+    this.deferredInputStream.detachSource();
+  }
+  clearUserTurn() {
+    var _a;
+    this.audioTranscript = "";
+    this.audioInterimTranscript = "";
+    this.audioPreflightTranscript = "";
+    this.finalTranscriptConfidence = [];
+    this.userTurnCommitted = false;
+    (_a = this.sttTask) == null ? void 0 : _a.cancelAndWait().finally(() => {
+      this.sttTask = import_utils.Task.from(({ signal }) => this.createSttTask(this.stt, signal));
+      this.sttTask.result.catch((err) => {
+        this.logger.error(`Error running STT task: ${err}`);
+      });
+    });
+  }
+  commitUserTurn(audioDetached) {
+    var _a;
+    const commitUserTurnTask = (delayDuration = 500) => async (controller) => {
+      if (Date.now() - this.lastFinalTranscriptTime > delayDuration) {
+        if (audioDetached && this.sampleRate !== void 0) {
+          const numSamples = Math.floor(this.sampleRate * 0.5);
+          const silence = new Int16Array(numSamples * 2);
+          const silenceFrame = new import_rtc_node.AudioFrame(silence, this.sampleRate, 1, numSamples);
+          this.silenceAudioWriter.write(silenceFrame);
+        }
+        await (0, import_utils.delay)(delayDuration, { signal: controller.signal });
+      }
+      if (this.audioInterimTranscript) {
+        this.audioTranscript = `${this.audioTranscript} ${this.audioInterimTranscript}`.trim();
+      }
+      this.audioInterimTranscript = "";
+      const chatCtx = this.hooks.retrieveChatCtx();
+      this.logger.debug("running EOU detection on commitUserTurn");
+      this.runEOUDetection(chatCtx);
+      this.userTurnCommitted = true;
+    };
+    (_a = this.commitUserTurnTask) == null ? void 0 : _a.cancel();
+    this.commitUserTurnTask = import_utils.Task.from(commitUserTurnTask());
+    this.commitUserTurnTask.result.then(() => {
+      this.logger.debug("User turn committed");
+    }).catch((err) => {
+      this.logger.error(err, "Error in user turn commit task:");
+    });
+  }
+  async close() {
+    var _a, _b, _c, _d, _e, _f;
+    this.detachInputAudioStream();
+    this.silenceAudioWriter.releaseLock();
+    await ((_a = this.commitUserTurnTask) == null ? void 0 : _a.cancelAndWait());
+    await ((_b = this.sttTask) == null ? void 0 : _b.cancelAndWait());
+    await ((_c = this.vadTask) == null ? void 0 : _c.cancelAndWait());
+    await ((_d = this.bounceEOUTask) == null ? void 0 : _d.cancelAndWait());
+    await ((_e = this.interruptionTask) == null ? void 0 : _e.cancelAndWait());
+    await ((_f = this.interruptionStream) == null ? void 0 : _f.close());
+  }
+  _endUserTurnSpan({
+    transcript,
+    confidence,
+    transcriptionDelay,
+    endOfUtteranceDelay
+  }) {
+    if (this.userTurnSpan) {
+      this.userTurnSpan.setAttributes({
+        [import_telemetry.traceTypes.ATTR_USER_TRANSCRIPT]: transcript,
+        [import_telemetry.traceTypes.ATTR_TRANSCRIPT_CONFIDENCE]: confidence,
+        [import_telemetry.traceTypes.ATTR_TRANSCRIPTION_DELAY]: transcriptionDelay,
+        [import_telemetry.traceTypes.ATTR_END_OF_TURN_DELAY]: endOfUtteranceDelay
+      });
+      this.userTurnSpan.end();
+      this.userTurnSpan = void 0;
+    }
+  }
+  get vadBaseTurnDetection() {
+    return ["vad", void 0].includes(this.turnDetectionMode);
+  }
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  AudioRecognition
+});
+//# sourceMappingURL=audio_recognition.cjs.map