@stream-io/video-client 1.42.3 → 1.43.0

package/dist/src/Call.d.ts

@@ -1,7 +1,7 @@
 import { Publisher, Subscriber, TrackPublishOptions } from './rtc';
 import { CallState } from './store';
 import { ScopedLogger } from './logger';
-import type { AcceptCallResponse, BlockUserResponse, CallRingEvent, CallSettingsResponse, CollectUserFeedbackRequest, CollectUserFeedbackResponse, DeleteCallRequest, DeleteCallResponse, EndCallResponse, GetCallReportResponse, GetCallResponse, GetCallSessionParticipantStatsDetailsResponse, GetOrCreateCallRequest, GetOrCreateCallResponse, GoLiveRequest, GoLiveResponse, JoinCallResponse, KickUserRequest, KickUserResponse, ListRecordingsResponse, ListTranscriptionsResponse, MuteUsersResponse, PinRequest, PinResponse, QueryCallMembersRequest, QueryCallMembersResponse, QueryCallSessionParticipantStatsResponse, QueryCallSessionParticipantStatsTimelineResponse, QueryCallStatsMapResponse, RejectCallResponse, RequestPermissionRequest, RequestPermissionResponse, RingCallRequest, RingCallResponse, SendCallEventResponse, SendReactionRequest, SendReactionResponse, StartClosedCaptionsRequest, StartClosedCaptionsResponse, StartFrameRecordingRequest, StartFrameRecordingResponse, StartHLSBroadcastingResponse, StartRTMPBroadcastsRequest, StartRTMPBroadcastsResponse, StartTranscriptionRequest, StartTranscriptionResponse, StopAllRTMPBroadcastsResponse, StopClosedCaptionsRequest, StopClosedCaptionsResponse, StopFrameRecordingResponse, StopHLSBroadcastingResponse, StopLiveRequest, StopLiveResponse, StopRecordingResponse, StopRTMPBroadcastsResponse, StopTranscriptionResponse, UnblockUserResponse, UnpinRequest, UnpinResponse, UpdateCallMembersRequest, UpdateCallMembersResponse, UpdateCallRequest, UpdateCallResponse, UpdateUserPermissionsRequest, UpdateUserPermissionsResponse } from './gen/coordinator';
+import type { AcceptCallResponse, BlockUserResponse, CallRingEvent, CallSettingsResponse, CollectUserFeedbackRequest, CollectUserFeedbackResponse, DeleteCallRequest, DeleteCallResponse, DeleteRecordingResponse, DeleteTranscriptionResponse, EndCallResponse, GetCallReportResponse, GetCallResponse, GetCallSessionParticipantStatsDetailsResponse, GetOrCreateCallRequest, GetOrCreateCallResponse, GoLiveRequest, GoLiveResponse, JoinCallResponse, KickUserRequest, KickUserResponse, ListRecordingsResponse, ListTranscriptionsResponse, MuteUsersResponse, PinRequest, PinResponse, QueryCallMembersRequest, QueryCallMembersResponse, QueryCallParticipantsRequest, QueryCallParticipantsResponse, QueryCallSessionParticipantStatsResponse, QueryCallSessionParticipantStatsTimelineResponse, QueryCallStatsMapResponse, RejectCallResponse, RequestPermissionRequest, RequestPermissionResponse, RingCallRequest, RingCallResponse, SendCallEventResponse, SendReactionRequest, SendReactionResponse, StartClosedCaptionsRequest, StartClosedCaptionsResponse, StartFrameRecordingRequest, StartFrameRecordingResponse, StartHLSBroadcastingResponse, StartRTMPBroadcastsRequest, StartRTMPBroadcastsResponse, StartTranscriptionRequest, StartTranscriptionResponse, StopAllRTMPBroadcastsResponse, StopClosedCaptionsRequest, StopClosedCaptionsResponse, StopFrameRecordingResponse, StopHLSBroadcastingResponse, StopLiveRequest, StopLiveResponse, StopRecordingResponse, StopRTMPBroadcastsResponse, StopTranscriptionResponse, UnblockUserResponse, UnpinRequest, UnpinResponse, UpdateCallMembersRequest, UpdateCallMembersResponse, UpdateCallRequest, UpdateCallResponse, UpdateUserPermissionsRequest, UpdateUserPermissionsResponse } from './gen/coordinator';
 import { AudioTrackType, CallConstructor, CallLeaveOptions, CallRecordingType, ClientPublishOptions, ClosedCaptionsSettings, JoinCallData, StartCallRecordingFnType, TrackMuteType, VideoTrackType } from './types';
 import { ClientCapability, TrackType, VideoDimension } from './gen/video/sfu/models/models';
 import { Tracer } from './stats';
@@ -631,6 +631,15 @@ export declare class Call {
      * @returns
      */
     queryMembers: (request?: Omit<QueryCallMembersRequest, "type" | "id">) => Promise<QueryCallMembersResponse>;
+    /**
+     * Query call participants with optional filters.
+     *
+     * @param data the request data.
+     * @param params optional query parameters.
+     */
+    queryParticipants: (data?: QueryCallParticipantsRequest, params?: {
+        limit?: number;
+    }) => Promise<QueryCallParticipantsResponse>;
     /**
      * Will update the call members.
      *
@@ -653,14 +662,45 @@ export declare class Call {
      * Otherwise, all recordings for the current call will be returned.
      *
      * @param callSessionId the call session id to retrieve recordings for.
+     * @deprecated use {@link listRecordings} instead.
      */
     queryRecordings: (callSessionId?: string) => Promise<ListRecordingsResponse>;
+    /**
+     * Retrieves the list of recordings for the current call or call session.
+     *
+     * If `callSessionId` is provided, it will return the recordings for that call session.
+     * Otherwise, all recordings for the current call will be returned.
+     *
+     * @param callSessionId the call session id to retrieve recordings for.
+     */
+    listRecordings: (callSessionId?: string) => Promise<ListRecordingsResponse>;
+    /**
+     * Deletes a recording for the given call session.
+     *
+     * @param callSessionId the call session id that the recording belongs to.
+     * @param filename the recording filename.
+     */
+    deleteRecording: (callSessionId: string, filename: string) => Promise<DeleteRecordingResponse>;
+    /**
+     * Deletes a transcription for the given call session.
+     *
+     * @param callSessionId the call session id that the transcription belongs to.
+     * @param filename the transcription filename.
+     */
+    deleteTranscription: (callSessionId: string, filename: string) => Promise<DeleteTranscriptionResponse>;
     /**
      * Retrieves the list of transcriptions for the current call.
      *
      * @returns the list of transcriptions.
+     * @deprecated use {@link listTranscriptions} instead.
      */
     queryTranscriptions: () => Promise<ListTranscriptionsResponse>;
+    /**
+     * Retrieves the list of transcriptions for the current call.
+     *
+     * @returns the list of transcriptions.
+     */
+    listTranscriptions: () => Promise<ListTranscriptionsResponse>;
     /**
      * Retrieve call statistics for a particular call session (historical).
      * Here `callSessionID` is mandatory.

package/dist/src/devices/MicrophoneManager.d.ts

@@ -10,6 +10,7 @@ export declare class MicrophoneManager extends AudioDeviceManager<MicrophoneMana
     private speakingWhileMutedNotificationEnabled;
     private soundDetectorConcurrencyTag;
     private soundDetectorCleanup?;
+    private soundDetectorDeviceId?;
     private noAudioDetectorCleanup?;
     private rnSpeechDetector;
     private noiseCancellation;
@@ -69,5 +70,6 @@ export declare class MicrophoneManager extends AudioDeviceManager<MicrophoneMana
     protected doSetAudioBitrateProfile(profile: AudioBitrateProfile): void;
     private startSpeakingWhileMutedDetection;
     private stopSpeakingWhileMutedDetection;
+    private teardownSpeakingWhileMutedDetection;
     private hasPermission;
 }

package/dist/src/helpers/no-audio-detector.d.ts

@@ -4,12 +4,6 @@ export type NoAudioDetectorOptions = {
      * Defaults to 350ms.
      */
     detectionFrequencyInMs?: number;
-    /**
-     * Defines the audio level threshold. Values below this are considered no audio.
-     * Defaults to 0. This value should be in the range of 0-255.
-     * Only applies to browser implementation.
-     */
-    audioLevelThreshold?: number;
     /**
      * Duration of continuous no-audio (in ms) before emitting the first event.
      */
@@ -23,7 +17,7 @@ export type NoAudioDetectorOptions = {
     /**
      * See https://developer.mozilla.org/en-US/docs/web/api/analysernode/fftsize
      *
-     * Defaults to 256.
+     * Defaults to 512.
      * Only applies to browser implementation.
      */
     fftSize?: number;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@stream-io/video-client",
-  "version": "1.42.3",
+  "version": "1.43.0",
   "main": "dist/index.cjs.js",
   "module": "dist/index.es.js",
   "browser": "dist/index.browser.es.js",

package/src/Call.ts

@@ -39,6 +39,8 @@ import type {
   Credentials,
   DeleteCallRequest,
   DeleteCallResponse,
+  DeleteRecordingResponse,
+  DeleteTranscriptionResponse,
   EndCallResponse,
   GetCallReportResponse,
   GetCallResponse,
@@ -59,6 +61,8 @@ import type {
   PinResponse,
   QueryCallMembersRequest,
   QueryCallMembersResponse,
+  QueryCallParticipantsRequest,
+  QueryCallParticipantsResponse,
   QueryCallSessionParticipantStatsResponse,
   QueryCallSessionParticipantStatsTimelineResponse,
   QueryCallStatsMapResponse,
@@ -1919,6 +1923,7 @@ export class Call {
         'Updating publish options after joining the call does not have an effect',
       );
     }
+    this.tracer.trace('updatePublishOptions', options);
     this.clientPublishOptions = { ...this.clientPublishOptions, ...options };
   };
 
@@ -2494,6 +2499,22 @@ export class Call {
     });
   };
 
+  /**
+   * Query call participants with optional filters.
+   *
+   * @param data the request data.
+   * @param params optional query parameters.
+   */
+  queryParticipants = async (
+    data: QueryCallParticipantsRequest = {},
+    params: { limit?: number } = {},
+  ): Promise<QueryCallParticipantsResponse> => {
+    return this.streamClient.post<
+      QueryCallParticipantsResponse,
+      QueryCallParticipantsRequest
+    >(`${this.streamClientBasePath}/participants`, data, params);
+  };
+
   /**
    * Will update the call members.
    *
@@ -2555,9 +2576,24 @@ export class Call {
    * Otherwise, all recordings for the current call will be returned.
    *
    * @param callSessionId the call session id to retrieve recordings for.
+   * @deprecated use {@link listRecordings} instead.
    */
   queryRecordings = async (
     callSessionId?: string,
+  ): Promise<ListRecordingsResponse> => {
+    return this.listRecordings(callSessionId);
+  };
+
+  /**
+   * Retrieves the list of recordings for the current call or call session.
+   *
+   * If `callSessionId` is provided, it will return the recordings for that call session.
+   * Otherwise, all recordings for the current call will be returned.
+   *
+   * @param callSessionId the call session id to retrieve recordings for.
+   */
+  listRecordings = async (
+    callSessionId?: string,
   ): Promise<ListRecordingsResponse> => {
     let endpoint = this.streamClientBasePath;
     if (callSessionId) {
@@ -2568,12 +2604,52 @@ export class Call {
     );
   };
 
+  /**
+   * Deletes a recording for the given call session.
+   *
+   * @param callSessionId the call session id that the recording belongs to.
+   * @param filename the recording filename.
+   */
+  deleteRecording = async (
+    callSessionId: string,
+    filename: string,
+  ): Promise<DeleteRecordingResponse> => {
+    return this.streamClient.delete<DeleteRecordingResponse>(
+      `${this.streamClientBasePath}/${encodeURIComponent(callSessionId)}/recordings/${encodeURIComponent(filename)}`,
+    );
+  };
+
+  /**
+   * Deletes a transcription for the given call session.
+   *
+   * @param callSessionId the call session id that the transcription belongs to.
+   * @param filename the transcription filename.
+   */
+  deleteTranscription = async (
+    callSessionId: string,
+    filename: string,
+  ): Promise<DeleteTranscriptionResponse> => {
+    return this.streamClient.delete<DeleteTranscriptionResponse>(
+      `${this.streamClientBasePath}/${encodeURIComponent(callSessionId)}/transcriptions/${encodeURIComponent(filename)}`,
+    );
+  };
+
   /**
    * Retrieves the list of transcriptions for the current call.
    *
    * @returns the list of transcriptions.
+   * @deprecated use {@link listTranscriptions} instead.
    */
   queryTranscriptions = async (): Promise<ListTranscriptionsResponse> => {
+    return this.listTranscriptions();
+  };
+
+  /**
+   * Retrieves the list of transcriptions for the current call.
+   *
+   * @returns the list of transcriptions.
+   */
+  listTranscriptions = async (): Promise<ListTranscriptionsResponse> => {
     return this.streamClient.get<ListTranscriptionsResponse>(
       `${this.streamClientBasePath}/transcriptions`,
     );

package/src/devices/MicrophoneManager.ts

@@ -34,6 +34,7 @@ export class MicrophoneManager extends AudioDeviceManager<MicrophoneManagerState
   private speakingWhileMutedNotificationEnabled = true;
   private soundDetectorConcurrencyTag = Symbol('soundDetectorConcurrencyTag');
   private soundDetectorCleanup?: () => Promise<void>;
+  private soundDetectorDeviceId?: string;
   private noAudioDetectorCleanup?: () => Promise<void>;
   private rnSpeechDetector: RNSpeechDetector | undefined;
   private noiseCancellation: INoiseCancellation | undefined;
@@ -385,6 +386,11 @@ export class MicrophoneManager extends AudioDeviceManager<MicrophoneManagerState
 
   private async startSpeakingWhileMutedDetection(deviceId?: string) {
     await withoutConcurrency(this.soundDetectorConcurrencyTag, async () => {
+      if (this.soundDetectorCleanup && this.soundDetectorDeviceId === deviceId)
+        return;
+
+      await this.teardownSpeakingWhileMutedDetection();
+
       if (isReactNative()) {
         this.rnSpeechDetector = new RNSpeechDetector();
         const unsubscribe = await this.rnSpeechDetector.start((event) => {
@@ -404,16 +410,26 @@ export class MicrophoneManager extends AudioDeviceManager<MicrophoneManagerState
           this.state.setSpeakingWhileMuted(event.isSoundDetected);
         });
       }
+
+      this.soundDetectorDeviceId = deviceId;
     });
   }
 
   private async stopSpeakingWhileMutedDetection() {
     await withoutConcurrency(this.soundDetectorConcurrencyTag, async () => {
-      if (!this.soundDetectorCleanup) return;
-      const soundDetectorCleanup = this.soundDetectorCleanup;
-      this.soundDetectorCleanup = undefined;
-      this.state.setSpeakingWhileMuted(false);
-      await soundDetectorCleanup();
+      return this.teardownSpeakingWhileMutedDetection();
+    });
+  }
+
+  private async teardownSpeakingWhileMutedDetection(): Promise<void> {
+    const soundDetectorCleanup = this.soundDetectorCleanup;
+    this.soundDetectorCleanup = undefined;
+    this.soundDetectorDeviceId = undefined;
+    this.state.setSpeakingWhileMuted(false);
+    if (!soundDetectorCleanup) return;
+
+    await soundDetectorCleanup().catch((err) => {
+      this.logger.warn('Failed to stop speaking while muted detector', err);
     });
   }
 

package/src/devices/__tests__/MicrophoneManager.test.ts

@@ -222,6 +222,38 @@ describe('MicrophoneManager', () => {
       }
     });
 
+    it('should not create duplicate sound detectors for the same device', async () => {
+      const detectorMock = vi.mocked(createSoundDetector);
+      const cleanup = vi.fn(async () => {});
+      detectorMock.mockImplementationOnce(() => cleanup);
+
+      await manager['startSpeakingWhileMutedDetection']('device-1');
+      await manager['startSpeakingWhileMutedDetection']('device-1');
+
+      expect(detectorMock).toHaveBeenCalledTimes(1);
+
+      await manager['stopSpeakingWhileMutedDetection']();
+      expect(cleanup).toHaveBeenCalledTimes(1);
+    });
+
+    it('should cleanup previous detector before starting a new device detector', async () => {
+      const detectorMock = vi.mocked(createSoundDetector);
+      const cleanupFirst = vi.fn(async () => {});
+      const cleanupSecond = vi.fn(async () => {});
+      detectorMock
+        .mockImplementationOnce(() => cleanupFirst)
+        .mockImplementationOnce(() => cleanupSecond);
+
+      await manager['startSpeakingWhileMutedDetection']('device-1');
+      await manager['startSpeakingWhileMutedDetection']('device-2');
+
+      expect(detectorMock).toHaveBeenCalledTimes(2);
+      expect(cleanupFirst).toHaveBeenCalledTimes(1);
+
+      await manager['stopSpeakingWhileMutedDetection']();
+      expect(cleanupSecond).toHaveBeenCalledTimes(1);
+    });
+
     // --- this ---
     it('should stop speaking while muted notifications if user loses permission to send audio', async () => {
       await manager.enable();
@@ -530,6 +562,26 @@ describe('MicrophoneManager', () => {
     });
   });
 
+  describe('no-audio detector configuration', () => {
+    it('applies silence threshold and emit interval in runtime monitoring', async () => {
+      const noAudioDetector = vi.mocked(createNoAudioDetector);
+
+      manager.setSilenceThreshold(3000);
+      manager['call'].state.setCallingState(CallingState.JOINED);
+      await manager.enable();
+
+      await vi.waitFor(() => {
+        expect(noAudioDetector).toHaveBeenCalled();
+      });
+
+      const options = noAudioDetector.mock.calls.at(-1)?.[1];
+      expect(options).toMatchObject({
+        noAudioThresholdMs: 3000,
+        emitIntervalMs: 3000,
+      });
+    });
+  });
+
   afterEach(() => {
     vi.restoreAllMocks();
     vi.clearAllMocks();

package/src/devices/__tests__/MicrophoneManagerRN.test.ts

@@ -16,6 +16,7 @@ import { SoundStateChangeHandler } from '../../helpers/sound-detector';
 import { settled, withoutConcurrency } from '../../helpers/concurrency';
 
 let handler: SoundStateChangeHandler = () => {};
+let unsubscribeHandlers: ReturnType<typeof vi.fn>[] = [];
 
 vi.mock('../../helpers/platforms.ts', () => {
   return {
@@ -51,7 +52,9 @@ vi.mock('../../helpers/RNSpeechDetector.ts', () => {
     RNSpeechDetector: vi.fn().mockImplementation(() => ({
       start: vi.fn((callback) => {
         handler = callback;
-        return vi.fn();
+        const unsubscribe = vi.fn();
+        unsubscribeHandlers.push(unsubscribe);
+        return unsubscribe;
       }),
       stop: vi.fn(),
       onSpeakingDetectedStateChange: vi.fn(),
@@ -63,6 +66,7 @@ describe('MicrophoneManager React Native', () => {
   let manager: MicrophoneManager;
   let checkPermissionMock: ReturnType<typeof vi.fn>;
   beforeEach(() => {
+    unsubscribeHandlers = [];
     checkPermissionMock = vi.fn(async () => true);
 
     globalThis.streamRNVideoSDK = {
@@ -153,6 +157,27 @@ describe('MicrophoneManager React Native', () => {
     expect(manager.state.speakingWhileMuted).toBe(false);
   });
 
+  it('should not create duplicate speech detectors for the same device', async () => {
+    await manager['startSpeakingWhileMutedDetection']('device-1');
+    await manager['startSpeakingWhileMutedDetection']('device-1');
+
+    expect(unsubscribeHandlers).toHaveLength(1);
+
+    await manager['stopSpeakingWhileMutedDetection']();
+    expect(unsubscribeHandlers[0]).toHaveBeenCalledTimes(1);
+  });
+
+  it('should cleanup previous speech detector before starting a new one', async () => {
+    await manager['startSpeakingWhileMutedDetection']('device-1');
+    await manager['startSpeakingWhileMutedDetection']('device-2');
+
+    expect(unsubscribeHandlers).toHaveLength(2);
+    expect(unsubscribeHandlers[0]).toHaveBeenCalledTimes(1);
+
+    await manager['stopSpeakingWhileMutedDetection']();
+    expect(unsubscribeHandlers[1]).toHaveBeenCalledTimes(1);
+  });
+
   it('should stop speaking while muted notifications if user loses permission to send audio', async () => {
     await manager.enable();
     await manager.disable();

package/src/devices/__tests__/web-audio.mocks.ts

@@ -18,8 +18,12 @@ export const createMockAnalyserNode = (
     get frequencyBinCount() {
       return fftSize / 2;
     },
-    // Default implementation fills array with zeros
-    // Tests can override with mockImplementation to simulate different audio levels
+    // Default implementation fills array with midpoint (silence waveform)
+    // Tests can override with mockImplementation to simulate different audio levels.
+    getByteTimeDomainData: vi.fn((array: Uint8Array) => {
+      array.fill(128);
+    }),
+    // Keep frequency-domain API for other helpers that use it.
     getByteFrequencyData: vi.fn((array: Uint8Array) => {
       array.fill(0);
     }),

package/src/helpers/__tests__/no-audio-detector.test.ts

@@ -12,7 +12,7 @@ describe('no-audio-detector (browser)', () => {
   let mockAudioContext: ReturnType<typeof setupAudioContextMock>;
   let audioStream: MediaStream;
   type MockAnalyserNode = ReturnType<typeof createMockAnalyserNode> & {
-    getByteFrequencyData: ReturnType<typeof vi.fn>;
+    getByteTimeDomainData: ReturnType<typeof vi.fn>;
   };
 
   const getAnalyserNode = () => {
@@ -30,7 +30,6 @@ describe('no-audio-detector (browser)', () => {
       noAudioThresholdMs: 5000,
       emitIntervalMs: 5000,
       detectionFrequencyInMs: 500,
-      audioLevelThreshold: 1, // Use threshold of 1 so level 0 is detected as "no audio"
       ...overrides,
     });
 
@@ -38,9 +37,19 @@ describe('no-audio-detector (browser)', () => {
   };
 
   const setAudioLevel = (analyserNode: MockAnalyserNode, level: number) => {
-    vi.mocked(analyserNode.getByteFrequencyData).mockImplementation((array) => {
-      array.fill(level);
-    });
+    const amplitude = Math.min(Math.max(level, 0), 127);
+    vi.mocked(analyserNode.getByteTimeDomainData).mockImplementation(
+      (array) => {
+        if (amplitude === 0) {
+          array.fill(128);
+          return;
+        }
+
+        for (let i = 0; i < array.length; i++) {
+          array[i] = i % 2 === 0 ? 128 + amplitude : 128 - amplitude;
+        }
+      },
+    );
   };
 
   beforeEach(() => {
@@ -92,6 +101,16 @@ describe('no-audio-detector (browser)', () => {
       expect(onCaptureStatusChange).toHaveBeenCalledWith(false);
     });
 
+    it('should treat tiny 127/128 jitter as no audio', () => {
+      const { onCaptureStatusChange, analyserNode } = createDetector();
+      setAudioLevel(analyserNode, 1);
+
+      vi.advanceTimersByTime(5500);
+
+      expect(onCaptureStatusChange).toHaveBeenCalledTimes(1);
+      expect(onCaptureStatusChange).toHaveBeenLastCalledWith(false);
+    });
+
     it('should respect custom emit interval', () => {
       const { onCaptureStatusChange, analyserNode } = createDetector({
         noAudioThresholdMs: 3000,
@@ -114,6 +133,27 @@ describe('no-audio-detector (browser)', () => {
   });
 
   describe('audio detection', () => {
+    it('should stop and emit audio detected when sound appears before threshold', () => {
+      const { onCaptureStatusChange, analyserNode } = createDetector();
+      setAudioLevel(analyserNode, 0);
+
+      // Start in no-audio detecting mode but stay below threshold.
+      vi.advanceTimersByTime(3000);
+      expect(onCaptureStatusChange).not.toHaveBeenCalled();
+
+      // Audio appears before no-audio threshold is reached.
+      setAudioLevel(analyserNode, 10);
+      vi.advanceTimersByTime(500);
+
+      expect(onCaptureStatusChange).toHaveBeenCalledTimes(1);
+      expect(onCaptureStatusChange).toHaveBeenLastCalledWith(true);
+
+      // Detector should be stopped.
+      onCaptureStatusChange.mockClear();
+      vi.advanceTimersByTime(10000);
+      expect(onCaptureStatusChange).not.toHaveBeenCalled();
+    });
+
     it('should stop checking after audio is detected', () => {
       const { onCaptureStatusChange, analyserNode } = createDetector();
       setAudioLevel(analyserNode, 0);
@@ -134,22 +174,6 @@ describe('no-audio-detector (browser)', () => {
       vi.advanceTimersByTime(10000);
       expect(onCaptureStatusChange).not.toHaveBeenCalled();
     });
-
-    it('should respect custom audio level threshold', () => {
-      const { onCaptureStatusChange, analyserNode } = createDetector({
-        audioLevelThreshold: 20, // Custom threshold
-      });
-      setAudioLevel(analyserNode, 15);
-
-      // Should detect as no audio since 15 < 20
-      vi.advanceTimersByTime(5500);
-      expect(onCaptureStatusChange).toHaveBeenCalledWith(false);
-
-      setAudioLevel(analyserNode, 25);
-      vi.advanceTimersByTime(500);
-
-      expect(onCaptureStatusChange).toHaveBeenCalledWith(true);
-    });
   });
 
   describe('track state handling', () => {
@@ -176,7 +200,7 @@ describe('no-audio-detector (browser)', () => {
       expect(onCaptureStatusChange).toHaveBeenCalledTimes(1);
     });
 
-    it('should reset state when track ends', () => {
+    it('should emit no-audio when track ends', () => {
       const [track] = audioStream.getAudioTracks() as Array<
         Omit<MediaStreamTrack, 'readyState'> & { readyState: string }
       >;
@@ -191,9 +215,10 @@ describe('no-audio-detector (browser)', () => {
       // Advance detection cycle
       vi.advanceTimersByTime(500);
 
-      // Should not emit (track ended)
+      // Ended track should be treated as no-audio and eventually emit.
       vi.advanceTimersByTime(5000);
-      expect(onCaptureStatusChange).not.toHaveBeenCalled();
+      expect(onCaptureStatusChange).toHaveBeenCalledTimes(1);
+      expect(onCaptureStatusChange).toHaveBeenCalledWith(false);
     });
   });
 
@@ -247,7 +272,7 @@ describe('no-audio-detector (browser)', () => {
   });
 
   describe('edge cases', () => {
-    it('should handle empty audio tracks array', () => {
+    it('should emit no-audio when stream has no audio tracks', async () => {
       const onCaptureStatusChange = vi.fn();
       const emptyStream = {
         getAudioTracks: () => [],
@@ -260,12 +285,13 @@ describe('no-audio-detector (browser)', () => {
         detectionFrequencyInMs: 500,
       });
 
-      // Should not crash or emit events
+      // Missing track should be treated as no-audio and eventually emit.
       vi.advanceTimersByTime(10000);
-      expect(onCaptureStatusChange).not.toHaveBeenCalled();
+      expect(onCaptureStatusChange).toHaveBeenCalledTimes(1);
+      expect(onCaptureStatusChange).toHaveBeenCalledWith(false);
 
       // Cleanup should work
-      expect(() => stop()).not.toThrow();
+      await expect(stop()).resolves.toBeUndefined();
     });
   });
 });

package/src/helpers/no-audio-detector.ts

@@ -6,12 +6,6 @@ export type NoAudioDetectorOptions = {
    * Defaults to 350ms.
    */
   detectionFrequencyInMs?: number;
-  /**
-   * Defines the audio level threshold. Values below this are considered no audio.
-   * Defaults to 0. This value should be in the range of 0-255.
-   * Only applies to browser implementation.
-   */
-  audioLevelThreshold?: number;
   /**
    * Duration of continuous no-audio (in ms) before emitting the first event.
    */
@@ -25,7 +19,7 @@ export type NoAudioDetectorOptions = {
   /**
    * See https://developer.mozilla.org/en-US/docs/web/api/analysernode/fftsize
    *
-   * Defaults to 256.
+   * Defaults to 512.
    * Only applies to browser implementation.
    */
   fftSize?: number;
@@ -52,12 +46,23 @@ type StateTransition =
   | { shouldEmit: true; nextState: DetectorState; capturesAudio: boolean };
 
 /**
- * Analyzes frequency data to determine if audio is being captured.
+ * Analyzes time-domain waveform data to determine if audio is being captured.
+ * Uses the waveform RMS around the 128 midpoint for robust silence detection.
  */
-const hasAudio = (analyser: AnalyserNode, threshold: number): boolean => {
-  const data = new Uint8Array(analyser.frequencyBinCount);
-  analyser.getByteFrequencyData(data);
-  return data.some((value) => value > threshold);
+const hasAudio = (analyser: AnalyserNode): boolean => {
+  const data = new Uint8Array(analyser.fftSize);
+  analyser.getByteTimeDomainData(data);
+
+  let squareSum = 0;
+  for (const sample of data) {
+    const centered = sample - 128;
+    // Ignore tiny quantization/jitter around midpoint (e.g. 127/128 samples).
+    const signal = Math.abs(centered) <= 1 ? 0 : centered;
+    squareSum += signal * signal;
+  }
+
+  const rms = Math.sqrt(squareSum / data.length);
+  return rms > 0;
 };
 
 /** Helper for "no event" transitions */
@@ -81,9 +86,9 @@ const transitionState = (
   options: NoAudioDetectorOptions,
 ): StateTransition => {
   if (audioDetected) {
-    return state.kind === 'IDLE' || state.kind === 'EMITTING'
-      ? emit(true, state)
-      : noEmit(state);
+    // Any observed audio means the microphone is capturing.
+    // Emit recovery/success and let the caller stop the detector.
+    return emit(true, { kind: 'IDLE' });
   }
 
   const { noAudioThresholdMs, emitIntervalMs } = options;
@@ -137,21 +142,21 @@ export const createNoAudioDetector = (
 ) => {
   const {
     detectionFrequencyInMs = 350,
-    audioLevelThreshold = 0,
-    fftSize = 256,
+    fftSize = 512,
     onCaptureStatusChange,
   } = options;
 
   let state: DetectorState = { kind: 'IDLE' };
   const { audioContext, analyser } = createAudioAnalyzer(audioStream, fftSize);
   const detectionIntervalId = setInterval(() => {
-    const [audioTrack] = audioStream.getAudioTracks();
-    if (!audioTrack?.enabled || audioTrack.readyState === 'ended') {
+    const [track] = audioStream.getAudioTracks();
+    if (track && !track.enabled) {
       state = { kind: 'IDLE' };
       return;
     }
 
-    const audioDetected = hasAudio(analyser, audioLevelThreshold);
+    // Missing or ended track is treated as no-audio to surface abrupt capture loss.
+    const audioDetected = track?.readyState === 'live' && hasAudio(analyser);
     const transition = transitionState(state, audioDetected, options);
 
     state = transition.nextState;