889a295c5c4e0c1cf344e49f2020917a410f64d1,nussl/core/augmentation.py,,time_stretch,#Any#Any#,80

Before Change

    audio_data = signal.audio_data

    for row in range(audio_data.shape[0]):
        audio_row = audio_data[row, :]
        if librosa.__version__ > "0.6.2":
            audio_row = np.asfortranarray(audio_row)
        stretched_source.append(librosa.effects.time_stretch(audio_row, stretch_factor))

After Change

    Returns:
        stretched_signal: A copy of the original audio_signal, with augmented sources. 
    
    if not np.issubdtype(type(stretch_factor), np.number) or stretch_factor <= 0:
        raise ValueError("stretch_factor must be a positve scalar")

    sample_rate = audio_signal.sample_rate
    stretched_audio_data = []
    audio_data = audio_signal.audio_data

    for audio_row in audio_signal.get_channels():
        stretched_audio_data.append(librosa.effects.time_stretch(audio_row, stretch_factor))
    stretched_audio_data = np.array(stretched_audio_data)
    stretched_signal = audio_signal.make_copy_with_audio_data(stretched_audio_data)
    stretched_signal.stft()

    return stretched_signal

Instances