Robust Principal Component Analysis

Huang, Po-Sen, et al. “Singing-voice separation from monaural recordings using robust principal component analysis.” 2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2012.

@inproceedings{huang2012singing,
  title={Singing-voice separation from monaural recordings using robust principal component analysis},
  author={Huang, Po-Sen and Chen, Scott Deeann and Smaragdis, Paris and Hasegawa-Johnson, Mark},
  booktitle={2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages={57--60},
  year={2012},
  organization={IEEE}
}

import nussl
import matplotlib.pyplot as plt
import time
import warnings
import numpy as np

warnings.filterwarnings("ignore")
start_time = time.time()

def visualize_and_embed(sources):
    plt.figure(figsize=(10, 6))
    plt.subplot(211)
    nussl.utils.visualize_sources_as_masks(sources,
        y_axis='mel', db_cutoff=-40, alpha_amount=2.0)
    plt.subplot(212)
    nussl.utils.visualize_sources_as_waveform(
        sources, show_legend=False)
    plt.show()
    nussl.play_utils.multitrack(sources)

audio_path = nussl.efz_utils.download_audio_file(
    'schoolboy_fascination_excerpt.wav')
audio_signal = nussl.AudioSignal(audio_path)

separator = nussl.separation.factorization.RPCA(audio_signal)
estimates = separator()

estimates = {
    'Low-rank source': estimates[0],
    'Sparse source': estimates[1]
}

visualize_and_embed(estimates)

Matching file found at /home/pseetharaman/.nussl/audio/schoolboy_fascination_excerpt.wav, skipping download.

end_time = time.time()
time_taken = end_time - start_time
print(f'Time taken: {time_taken:.4f} seconds')

Time taken: 206.3886 seconds