AdaptiveVAD - Detects speech in audio frames, adapting to environmental noise.

Hierarchy

  • EventEmitter
    • AdaptiveVAD

Constructors

  • Creates a new AdaptiveVAD instance.

    Parameters

    • config: AdaptiveVADConfig = {}

      VAD configuration options.

    • calibrator: EnvironmentalCalibrator

      Instance of EnvironmentalCalibrator for noise profiles.

    • frameDurationMs: number

      Duration of each audio frame in milliseconds that will be processed. (e.g., for 16000Hz and 320 samples/frame, duration is 20ms).

    Returns AdaptiveVAD

Methods

  • Type Parameters

    • U extends keyof VADEmitterEvents

    Parameters

    • event: U
    • listener: VADEmitterEvents[U]

    Returns this

  • Type Parameters

    • U extends keyof VADEmitterEvents

    Parameters

    • event: U
    • Rest ...args: Parameters<VADEmitterEvents[U]>

    Returns boolean

  • Processes an incoming audio frame to detect voice activity.

    Parameters

    • audioFrame: Float32Array<ArrayBufferLike>

      A chunk of raw audio data (PCM).

    Returns VADResult

    The result of VAD processing for this frame.

  • Resets the VAD's internal state. Useful when starting a new audio stream or after a manual interruption.

    Returns void

  • Gets the current VAD state.

    Returns {
        isSpeaking: boolean;
        speechThreshold: number;
        silenceThreshold: number;
        consecutiveSpeechFrames: number;
        consecutiveSilenceFrames: number;
    }

    • isSpeaking: boolean
    • speechThreshold: number
    • silenceThreshold: number
    • consecutiveSpeechFrames: number
    • consecutiveSilenceFrames: number