mp.tasks.audio.AudioTranscriber

Performs streaming and unary speech-to-text transcription.

mp.tasks.audio.AudioTranscriber(
    handle: ctypes.c_void_p
)

handle Opaque C pointer created by MpAudioTranscriberCreate.

Methods

close

View source

close() -> None

Closes the transcriber and releases native C++ resources.

create_from_model_path

View source

@classmethod
create_from_model_path(
    model_path: str,
    delegate: mp.tasks.BaseOptions.Delegate = _BaseOptions.Delegate.CPU,
    sample_rate_hz: int = 16000
) -> 'AudioTranscriber'

Creates an AudioTranscriber from a .litertlm model path.

create_from_options

View source

@classmethod
create_from_options(
    options: mp.tasks.audio.AudioTranscriberOptions
) -> 'AudioTranscriber'

Creates an AudioTranscriber instance from AudioTranscriberOptions.

finish_stream

View source

finish_stream() -> mp.tasks.audio.AudioTranscriberResult

Flushes any buffered audio and finalizes the current stream.

prewarm

View source

prewarm() -> None

Runs a dummy warm-up pass to initialize execution kernels.

transcribe

View source

transcribe(
    audio_samples: Union[Sequence[float], np.ndarray],
    end_of_stream: bool = True
) -> mp.tasks.audio.AudioTranscriberResult

Transcribes mono PCM float32 audio samples in [-1.0, 1.0].

Args
audio_samples Sequence or 1D NumPy array of mono float32 PCM samples.
end_of_stream Whether this audio chunk concludes the current stream.

Returns
An AudioTranscriberResult containing recognized segments.

__enter__

View source

__enter__() -> 'AudioTranscriber'

__exit__

View source

__exit__(
    exc_type, exc_val, exc_tb
) -> None