AI による音楽ソース分離

pip install torch torchaudio matplotlib

import torch
import torchaudio
import matplotlib.pyplot as plt

from IPython.display import Audio
from torchaudio.pipelines import HDEMUCS_HIGH_MUSDB_PLUS
from torchaudio.utils import _download_asset

print(torch.__version__)
print(torchaudio.__version__)

bundle = HDEMUCS_HIGH_MUSDB_PLUS
model = bundle.get_model()

device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
model.to(device)

sample_rate = bundle.sample_rate

print(f"Sample rate: {sample_rate}")

from torchaudio.transforms import Fade


def separate_sources(
    model,
    mix,
    segment=10.0,
    overlap=0.1,
    device=None,
):
    """
    Perform music source separation using chunked inference.

    Args:
        segment: Length of each audio chunk in seconds
        overlap: Overlap ratio between consecutive chunks
        device: Inference device
    """

    if device is None:
        device = mix.device
    else:
        device = torch.device(device)

    batch, channels, length = mix.shape

    chunk_len = int(sample_rate * segment * (1 + overlap))

    start = 0
    end = chunk_len

    overlap_frames = overlap * sample_rate

    fade = Fade(
        fade_in_len=0,
        fade_out_len=int(overlap_frames),
        fade_shape="linear",
    )

    final = torch.zeros(
        batch,
        len(model.sources),
        channels,
        length,
        device=device,
    )

    while start < length - overlap_frames:

        chunk = mix[:, :, start:end]

        with torch.no_grad():
            out = model.forward(chunk)

        out = fade(out)

        final[:, :, :, start:end] += out

        if start == 0:
            fade.fade_in_len = int(overlap_frames)
            start += int(chunk_len - overlap_frames)
        else:
            start += chunk_len

        end += chunk_len

        if end >= length:
            fade.fade_out_len = 0

    return final

def plot_spectrogram(stft, title="Spectrogram"):
    magnitude = stft.abs()

    spectrogram = 20 * torch.log10(
        magnitude + 1e-8
    ).numpy()

    _, axis = plt.subplots(1, 1)

    axis.imshow(
        spectrogram,
        cmap="viridis",
        vmin=-60,
        vmax=0,
        origin="lower",
        aspect="auto",
    )

    axis.set_title(title)

    plt.tight_layout()

SAMPLE_SONG = _download_asset(
    "tutorial-assets/hdemucs_mix.wav"
)

waveform, sample_rate = torchaudio.load(SAMPLE_SONG)

waveform = waveform.to(device)

mixture = waveform

segment = 10
overlap = 0.1

ref = waveform.mean(0)

waveform = (
    waveform - ref.mean()
) / ref.std()

sources = separate_sources(
    model,
    waveform[None],
    device=device,
    segment=segment,
    overlap=overlap,
)[0]

sources = sources * ref.std() + ref.mean()

sources_list = model.sources

sources = list(sources)

audios = dict(zip(sources_list, sources))

N_FFT = 4096
N_HOP = 4

stft = torchaudio.transforms.Spectrogram(
    n_fft=N_FFT,
    hop_length=N_HOP,
    power=None,
)

segment_start = 150
segment_end = 155

frame_start = segment_start * sample_rate
frame_end = segment_end * sample_rate

drums_original = _download_asset(
    "tutorial-assets/hdemucs_drums_segment.wav"
)

bass_original = _download_asset(
    "tutorial-assets/hdemucs_bass_segment.wav"
)

vocals_original = _download_asset(
    "tutorial-assets/hdemucs_vocals_segment.wav"
)

other_original = _download_asset(
    "tutorial-assets/hdemucs_other_segment.wav"
)

drums_spec = audios["drums"][:, frame_start:frame_end].cpu()
bass_spec = audios["bass"][:, frame_start:frame_end].cpu()
vocals_spec = audios["vocals"][:, frame_start:frame_end].cpu()
other_spec = audios["other"][:, frame_start:frame_end].cpu()

mix_spec = mixture[:, frame_start:frame_end].cpu()

def output_results(
    original_source: torch.Tensor,
    predicted_source: torch.Tensor,
    source: str,
):

    plot_spectrogram(
        stft(predicted_source)[0],
        f"Spectrogram - {source}"
    )

    return Audio(
        predicted_source,
        rate=sample_rate
    )

plot_spectrogram(
    stft(mix_spec)[0],
    "Spectrogram - Mixture"
)

Audio(mix_spec, rate=sample_rate)

output_results(
    drums,
    drums_spec,
    "drums"
)

output_results(
    bass,
    bass_spec,
    "bass"
)

output_results(
    vocals,
    vocals_spec,
    "vocals"
)

output_results(
    other,
    other_spec,
    "other"
)

# Original mixture
# Audio(mixture, rate=sample_rate)

# Drums
# Audio(audios["drums"], rate=sample_rate)

# Bass
# Audio(audios["bass"], rate=sample_rate)

# Vocals
# Audio(audios["vocals"], rate=sample_rate)

# Other accompaniment
# Audio(audios["other"], rate=sample_rate)

Défossez, A. (2021).
Hybrid Spectrogram and Waveform Source Separation.
arXiv preprint arXiv:2111.03600.

Défossez, A., Usunier, N., Bottou, L., & Bach, F. (2019).
Music Source Separation in the Waveform Domain.
Proceedings of the 20th International Society for Music Information Retrieval Conference (ISMIR).

Rafii, Z., Liutkus, A., Stöter, F.-R., Mimilakis, S. I.,
Bittner, R., & Pardo, B. (2017).
The MUSDB18 Corpus for Music Separation.

PyTorch Audio Team.
Torchaudio Documentation.

PyTorch Audio Team.
Music Source Separation with Hybrid Demucs.

Stöter, F.-R., Liutkus, A., & Ito, N. (2019).
The 2018 Signal Separation Evaluation Campaign.
Journal of Open Source Software.

Jansson, A., Humphrey, E., Montecchio, N.,
Bittner, R., Kumar, A., & Weyde, T. (2017).
Singing Voice Separation with Deep U-Net Convolutional Networks.

Vincent, E., Gribonval, R., & Févotte, C. (2006).
Performance Measurement in Blind Audio Source Separation.
IEEE Transactions on Audio, Speech, and Language Processing.

Kellman, M., et al.
Deep Learning for Audio Signal Processing.

Müller, M. (2015).
Fundamentals of Music Processing.
Springer.