Interface AudioActivityDetector

Rolling speech-versus-music detector over a live PCM stream.

interface AudioActivityDetector {
    classifierId: string;
    options: ResolvedAudioActivityDetectorOptions;
    state: AudioActivityState;
    push(pcm: Buffer, sampleRate: number): AudioActivityUpdate;
    noteProviderMusicEvent(active: boolean, confidence: number): void;
    lastScores(): AudioActivityScores;
    reset(): void;
}

Index

Properties

classifierId options state

Methods

push noteProviderMusicEvent lastScores reset

Properties

`Readonly`classifierId

classifierId: string

Identifier of the underlying classifier.

`Readonly`options

options: ResolvedAudioActivityDetectorOptions

Settings actually in effect, after defaults and threshold reconciliation.

`Readonly`state

state: AudioActivityState

Current committed state.

Methods

push

push(pcm: Buffer, sampleRate: number): AudioActivityUpdate
Feeds a PCM chunk and advances the state machine.
Parameters
- pcm: Buffer
  Raw PCM16 LE mono bytes.
- sampleRate: number
  Sample rate of pcm in Hz.
Returns AudioActivityUpdate
Current state, whether it changed, and the latest scores.
- Defined in lib/translation/audio-activity/detector.ts:142

noteProviderMusicEvent

noteProviderMusicEvent(active: boolean, confidence: number): void
Records a music event reported by the upstream ASR provider.
Parameters
- active: boolean
  True when a music event started, false when it ended.
- confidence: number
  Provider confidence in [0, 1].
Returns void
- Defined in lib/translation/audio-activity/detector.ts:148

lastScores

lastScores(): AudioActivityScores
Most recent classification, or null when none has run.

Returns AudioActivityScores
- Defined in lib/translation/audio-activity/detector.ts:150

reset

reset(): void
Clears buffered audio and returns to the speech state.

Returns void
- Defined in lib/translation/audio-activity/detector.ts:152