Low-Level API

Direct session access for advanced use cases like custom audio processing pipelines.

Recording session

val session = SystemAudioSystem.createRecordingSession()

session.start()
delay(5.seconds)
session.stop()

val audioFlow = session.audioFlow.value
val recording = AudioRecording.fromAudioFlow(audioFlow!!)

You can request a specific audio format. The platform will try to honor it, falling back to a supported format if necessary:

val session = SystemAudioSystem.createRecordingSession(
    requestedFormat = AudioFormat(
        sampleRate = 16000,
        channels = Channels.Mono,
        encoding = SampleEncoding.PcmInt(IntBitDepth.Sixteen)
    )
)

Playback session

val session = SystemAudioSystem.createPlaybackSession()

session.load(audioFlow)
session.play()
session.state.first { it is AudioPlaybackSession.State.Finished }
session.stop()

Live processing

Process audio in real-time while recording:

val session = SystemAudioSystem.createRecordingSession()
session.start()

session.audioFlow.value?.collect { chunk ->
    val amplitude = calculateRMS(chunk)
    updateWaveform(amplitude)
}

AudioFlow

val audioFlow = AudioFlow(
    format = AudioFormat(
        sampleRate = 44100,
        channels = Channels.Mono,
        encoding = SampleEncoding.PcmInt(IntBitDepth.Sixteen)
    ),
    data = flow { emit(chunk) }
)

SystemAudioSystem

val inputs = SystemAudioSystem.listInputDevices()
val outputs = SystemAudioSystem.listOutputDevices()
val recordingSession = SystemAudioSystem.createRecordingSession(
    requestedDevice = device,
    requestedFormat = AudioQuality.Voice.format
)
val playbackSession = SystemAudioSystem.createPlaybackSession(device)

Last modified: 04 April 2026