AudioCapture/dev.piotrprus.audiocapture/CaptureConfig

CaptureConfig

data class CaptureConfig(val sampleRate: Int, val channels: Int = 1, val chunkDuration: Duration = 100.milliseconds, val pcm: PcmOutput? = PcmOutput(), val file: FileOutput? = null, val device: InputDevice? = null, val voiceProcessing: VoiceProcessing? = null, val interruption: InterruptionMode = InterruptionMode.Pause, val android: AndroidOptions = AndroidOptions(), val ios: IosOptions = IosOptions())(source)

What to capture and where it goes.

The defaults suit speech: 16 kHz mono PCM16 in 100 ms chunks, the format most speech-to-text APIs expect. Set file to also record, or set pcm to null to only record.

Constructors

CaptureConfig

constructor(sampleRate: Int, channels: Int = 1, chunkDuration: Duration = 100.milliseconds, pcm: PcmOutput? = PcmOutput(), file: FileOutput? = null, device: InputDevice? = null, voiceProcessing: VoiceProcessing? = null, interruption: InterruptionMode = InterruptionMode.Pause, android: AndroidOptions = AndroidOptions(), ios: IosOptions = IosOptions())

Properties

android

val android: AndroidOptions

channels

val channels: Int

1 (mono) or 2 (stereo). Whether stereo is two real microphones or one duplicated depends on the device, the input and (on Android) the audio source; AndroidAudioSource.Camcorder is the most likely to give real stereo.

chunkDuration

val chunkDuration: Duration

How much audio each AudioChunk holds and how often CaptureSession.level updates.

device

val device: InputDevice?

Preferred microphone, from AudioCapture.inputDevices. null is the system default.

file

val file: FileOutput?

A file written while capturing, or null for none.

interruption

val interruption: InterruptionMode

ios

val ios: IosOptions

pcm

val pcm: PcmOutput?

Raw PCM delivered through CaptureSession.chunks, or null for none.

sampleRate

val sampleRate: Int

Samples per second, per channel. 16 000, 44 100 and 48 000 work on all current phones; Android only guarantees 44 100, and AudioCapture.start throws if the device refuses a rate. AAC files are limited to 8 000–48 000.

voiceProcessing

val voiceProcessing: VoiceProcessing?

Echo cancellation, noise suppression and gain control, or null for the unprocessed signal. Experimental: see ExperimentalVoiceProcessing.