def process_directory(dir, n_rate): signal = [] for j, audioname in enumerate(f[dir]): holder_signal, sr = Audio.read(f'{origin_path}/{dir}/{audioname}', sr=n_rate) signal.extend(Audio.trim(holder_signal, 20)) signal = array(signal) Audio.write(f'{dest_path}/{n_rate}/{dir}.wav', signal, n_rate)
warped_masked_spectrogram = spec_augment_tensorflow.spec_augment(mfcc) warped_masked_spectrogram = warped_masked_spectrogram.numpy() # %% plt.figure(figsize=(10, 4)) librosa.display.specshow(librosa.power_to_db(warped_masked_spectrogram[0, :, :, 0], ref=np.max), y_axis='mel', fmax=8000, x_axis='time') plt.tight_layout() plt.title('SpecAugmented') plt.show() plt.close() audio_signal = librosa.core.spectrum.griffinlim( warped_masked_spectrogram[0, :, :, 0]) Audio.write('test/warped_audio.wav', audio_signal, sr) # %% plt.figure(figsize=(10, 4)) librosa.display.specshow(librosa.power_to_db(mfcc[0, :, :, 0], ref=np.max), y_axis='mel', fmax=8000, x_axis='time') plt.tight_layout() plt.title('MFCC') plt.show() plt.close() audio_signal = librosa.core.spectrum.griffinlim(mfcc[0, :, :, 0]) Audio.write('test/mfcc_audio.wav', audio_signal, sr)