MediaPipeTasksAudio Framework Reference

AudioTranscriber

class AudioTranscriber : NSObject

Performs automatic speech recognition (ASR) on streaming mono PCM float audio samples.

• init(options:)

  Creates a new instance of AudioTranscriber from the given AudioTranscriberOptions.

  Declaration

  Swift

  init(options: AudioTranscriberOptions) throws

• init(modelPath:)

  Creates a new instance of AudioTranscriber from a local model asset path.

  Declaration

  Swift

  convenience init(modelPath: String) throws

• transcribe(samples:endOfStream:)

  Transcribes an array of mono PCM float audio samples in [-1.0, 1.0].

  Declaration

  Swift

  func transcribe(samples: [NSNumber], endOfStream: Bool) throws -> AudioTranscriberResult

• finishStream()

  Flushes any remaining buffered audio and finalizes the current stream.

  Declaration

  Swift

  func finishStream() throws -> AudioTranscriberResult

• prewarm()

  Prewarms the underlying inference engine to reduce first-chunk latency.

  Declaration

  Swift

  func prewarm() throws

• close()

  Releases underlying inference engine and processor resources.

  Declaration

  Swift

  func close() throws