get_data_to_buffer with f0 processing

get_data_to_buffer

def get_data_to_buffer(train_config):
    buffer = list()
    text = process_text(train_config.data_path)
    audio_files = sorted(Path(train_config.audio_path).iterdir())
    hop_length = 256
    frame_period = hop_length / 22_050 * 1000

    start = time.perf_counter()
    for i, file in tqdm(zip(range(len(text)), audio_files)):

        mel_gt_name = os.path.join(
            train_config.mel_ground_truth, "ljspeech-mel-%05d.npy" % (i+1))
        mel_gt_target = np.load(mel_gt_name)
        duration = np.load(os.path.join(
            train_config.alignment_path, str(i)+".npy"))
        character = text[i][0:len(text[i])-1]
        character = np.array(
            text_to_sequence(character, train_config.text_cleaners))
        
        wav, sr = librosa.load(file, sr=None, dtype=np.float64)
        f0, t = pw.dio(wav, sr, frame_period=frame_period)
        assert f0.shape[0] == mel_gt_target.shape[0]

        character = torch.from_numpy(character)
        duration = torch.from_numpy(duration)
        mel_gt_target = torch.from_numpy(mel_gt_target)
        f0 = torch.from_numpy(f0)

        buffer.append({"text": character, "duration": duration,
                       "mel_target": mel_gt_target, "f0": f0})

    end = time.perf_counter()
    print("cost {:.2f}s to load all data into buffer.".format(end-start))

    return buffer