The Journal of the Acoustical Society of Korea (한국음향학회지)

The Acoustical Society of Korea (한국음향학회)
권호 표지
DOI QR코드

DOI QR Code

A study on combination of loss functions for effective mask-based speech enhancement in noisy environments

잡음 환경에 효과적인 마스크 기반 음성 향상을 위한 손실함수 조합에 관한 연구

  • 정재희 (인천대학교 컴퓨터공학부) ;
  • 김우일 (인천대학교 컴퓨터공학부)
  • Received : 2021.03.19
  • Accepted : 2021.04.22
  • Published : 2021.05.31
Download PDF
⟨ Previous Next ⟩

Abstract

In this paper, the mask-based speech enhancement is improved for effective speech recognition in noise environments. In the mask-based speech enhancement, enhanced spectrum is obtained by multiplying the noisy speech spectrum by the mask. The VoiceFilter (VF) model is used as the mask estimation, and the Spectrogram Inpainting (SI) technique is used to remove residual noise of enhanced spectrum. In this paper, we propose a combined loss to further improve speech enhancement. In order to effectively remove the residual noise in the speech, the positive part of the Triplet loss is used with the component loss. For the experiment TIMIT database is re-constructed using NOISEX92 noise and background music samples with various Signal to Noise Ratio (SNR) conditions. Source to Distortion Ratio (SDR), Perceptual Evaluation of Speech Quality (PESQ), and Short-Time Objective Intelligibility (STOI) are used as the metrics of performance evaluation. When the VF was trained with the mean squared error and the SI model was trained with the combined loss, SDR, PESQ, and STOI were improved by 0.5, 0.06, and 0.002 respectively compared to the system trained only with the mean squared error.

본 논문에서는 잡음 환경에서 효과적인 음성 인식을 위해 마스크 기반의 음성 향상 기법을 개선한다. 마스크 기반의 음성 향상 기법에서는 심층 신경망을 기반으로 추정한 마스크를 잡음 오염 음성에 곱하여 향상된 음성을 얻는다. 마스크 추정 모델로 VoiceFilter(VF) 모델을 사용하고 추정된 마스크로 얻은 음성으로부터 잔여 잡음을 보다 확실히 제거하기 위해 Spectrogram Inpainting(SI)기법을 적용한다. 본 논문에서는 음성 향상 결과를 보다 개선하기 위해 마스크 추정을 위한 모델 학습 과정에 사용되는 조합된 손실함수를 제안한다. 음성 구간에 남아 있는 잡음을 보다 효과적으로 제거하기 위해 잡음 오염 음성에 마스크를 적용한 Triplet 손실함수의 Positive 부분을 컴포넌트 손실함수와 조합하여 사용한다. 실험 평가를 위한 잡음 음성 데이터는 TIMIT 데이터베이스와 NOISEX92, 배경음악 잡음을 다양한 Signal to Noise Ratio(SNR) 조건으로 합성하여 만들어 사용한다. 음성 향상의 성능 평가는 Source to Distortion Ratio(SDR), Perceptual Evaluation of Speech Quality(PESQ), Short-Time Objective Intelligibility(STOI)를 이용한다. 실험을 통해 평균 제곱 오차로만 훈련된 기존 시스템과 비교하여, VF 모델은 평균 제곱 오차로 훈련하고 SI 모델은 조합된 손실함수를 사용하였을 때 SDR은 평균 0.5dB, PESQ는 평균 0.06, STOI는 평균 0.002만큼 성능이 향상된 것을 확인했다.

Keywords

Acknowledgement

이 논문은 정부(과학기술정보통신부)의 재원으로 한국연구재단의 지원을 받아 수행된 연구임(No. 2019R1F1A106299513).

References

  1. J. Lim and A. Oppenheim. "All-pole modeling of degraded speech," IEEE Trans. on Acoustics, Speech, and Signal Process. 26, 197-210 (1978). https://doi.org/10.1109/TASSP.1978.1163086
  2. S. F. Boll, "Suppression of acoustic noise in speech using spectral subtraction," IEEE Trans. on Acoustics, Speech and Signal Proccess. 27, 113-120 (1979). https://doi.org/10.1109/TASSP.1979.1163209
  3. Y. Ephraim and D. Malah, "Speech enhancement using minimum mean square error short time spectral amplitude estimator," IEEE Trans. on Acoustics, Speech and Signal Process. 32, 1109-1121 (1984). https://doi.org/10.1109/TASSP.1984.1164453
  4. R. Martin, "Spectral subtraction based on minimum statistics," Proc. EUSIPCO. 1182-1185 (1994).
  5. P. J. Moreno, B. Raj, and R. M. Stern, "Data-driven environmental compensation for speech recognition: a unified approach," Speech Communication, 24, 267-285 (1998). https://doi.org/10.1016/S0167-6393(98)00025-9
  6. W. Kim and J. H. L. Hansen, "Feature compensation in the cepstral domain employing model combination," Speech Communication, 51, 83-96 (2009). https://doi.org/10.1016/j.specom.2008.06.004
  7. J. Du, L.-R. Dai, and Q. Huo, "Synthesized stereo mapping via deep neural networks for noisy speech recognition," Proc. ICASSP. 1764-1768 (2014).
  8. K. Han, Y. He, D. Bagchi, E. F. -Luissier, and D. L. Wang, "Deep neural network based spectral feature mapping for robust speech recognition," Proc. Interspeech, 2484-2488 (2015).
  9. K. Tan and D. L. Wang, "A convolutional recurrent neural network for real-time speech enhancement," Proc. Interspeech, 3229-3233 (2018).
  10. Y. Xu, J. Du; L.-R. Dai, and C.-H. Lee, "A regression approach to speech enhancement based on deep neural networks," IEEE/ACM. Trans. on Audio, Speech, and Lang. Process. 23, 7-19 (2014).
  11. Q. Wang, H. Muckenhirn, K. Wilson, P. Sridhar, Z. Wu, J. Hershey, R. A. Saurous, R. J. Weiss, Y. Jia, and I. L. Moreno, "Voicefilter: Targeted voice separation by speaker-conditioned spectrogram masking," arXiv preprint arXiv:1810.04826 (2018).
  12. X. Hao, X. Su, S. Wen, Z. Wang, Y. Pan, F. Bao, and W. Chen, "Masking and inpainting: A two-stage speech enhancement approach for low SNR and non-stationary noise," Proc. ICASSP. 6959-6963 (2020).
  13. Z. Xu, S. Elshamy, and T. Fingscheidt, "Using separate losses for speech and noise in mask-based speech enhancement," Proc. ICASSP. 7519-7523 (2020).
  14. F. Schroff, D. Kalenichenko, and J. Philbin, "Facenet: A unified embedding for face recognition and clustering," Proc. IEEE conference on CVPR. 815-823 (2015).
  15. J. S. Garofolo, L. F. Lamel, W. M. Fisher, J. G. Fiscus, D. S. Pallett, and N. L. Dahlgren, "DARPA TIMIT acoustic-phonetic continous speech corpus CD-ROM. NIST speech disc 1-1.1," NASA STI/Recon Tech. Rep., 1993.
  16. A. Varga and H. J. M. Steeneken, "Assessment for automatic speech recognition: II. NOISEX-92: A database and an experiment to study the effect of additive noise on speech recognition systems," Speech Communication, 12, 247-251 (1993). https://doi.org/10.1016/0167-6393(93)90095-3
  17. E. Vincent, R. Gribonval, and C. Fevotte, "Performance measurement in blind audio source separation," IEEE Trans. on audio, speech, and lang. process. 14, 1462-1469 (2006). https://doi.org/10.1109/TSA.2005.858005
  18. A. W. Rix, J. G. Beerends, M. P. Hollier, and A. P. Hekstra, "Perceptual evaluation of speech quality (PESQ)-a new method for speech quality assessment of telephone networks and codecs," Proc. IEEE ICASSP. 01CH37221 (2001).
  19. C. H. Taal, R. C. Hendriks, R, Heusdens, and J. Jensen, "A short-time objective intelligibility measure for time- frequency weighted noisy speech," Proc. IEEE ICASSP. 4214-4217 (2010).