MediaPipeTasksAudio Framework Reference

TTSStream

final class TTSStream: Sendable

Bidirectional / pull-based streaming synthesis handle supporting Level 2 incremental LLM text pushing (pushText + finishInput) alongside readNext() and cancel().

• cancel()

  Lock-free cancellation for this stream.

  Declaration

  Swift

  final func cancel()

• close()

  Cancels any in-flight synthesis and marks this stream as closed.

  The underlying native MpTextToSpeechStreamPtr is destroyed in deinit once all in-flight calls on self have returned.

  Declaration

  Swift

  final func close()

• finishInput()

  Marks the end of incremental text input, flushing any trailing partial sentence.

  Declaration

  Swift

  final func finishInput() throws

• pushText(_:)

  Appends an incremental text fragment into the sentence chunker.

  Declaration

  Swift

  final func pushText(_ textDelta: String) throws

• readNext()

  Pulls the next synthesized AudioChunk synchronously on the calling thread, or returns nil when the stream reaches end-of-stream.

  Declaration

  Swift

  final func readNext() throws -> AudioChunk?