def audio_summary()

in python/mxboard/summary.py [0:0]

• 25 lines of code
• 4 McCabe index (conditional complexity)

def audio_summary(tag, audio, sample_rate=44100):
    """Outputs a `Summary` protocol buffer with audio data.

    Parameters
    ----------
        tag : str
            A name for the generated summary. Will also serve as a series name in TensorBoard.
        audio : MXNet `NDArray` or `numpy.ndarray`
            Audio data that can be squeezed into 1D array. The values are in the range [-1, 1].
        sample_rate : int
            Sampling frequency. 44,100Hz is a common sampling frequency.

    Returns
    -------
        A `Summary` protobuf of the audio data.
    """
    audio = audio.squeeze()
    if audio.ndim != 1:
        raise ValueError('input audio must be squeezable to 1D, input audio squeezed '
                         'shape is {}'.format(audio.shape))
    audio = _make_numpy_array(audio)
    tensor_list = [int(32767.0 * x) for x in audio]
    fio = io.BytesIO()
    wave_writer = wave.open(fio, 'wb')
    wave_writer.setnchannels(1)
    wave_writer.setsampwidth(2)
    wave_writer.setframerate(sample_rate)
    tensor_enc = b''
    for v in tensor_list:  # pylint: disable=invalid-name
        tensor_enc += struct.pack('<h', v)
    wave_writer.writeframes(tensor_enc)
    wave_writer.close()
    audio_string = fio.getvalue()
    fio.close()
    audio = Summary.Audio(sample_rate=sample_rate,
                          num_channels=1,
                          length_frames=len(tensor_list),
                          encoded_audio_string=audio_string,
                          content_type='audio/wav')
    return Summary(value=[Summary.Value(tag=tag, audio=audio)])