A Supervised Speech Enhancement Approach with Residual Noise Control for Voice Communication
Source abstract
Abstract
For voice communication, it is important to extract the speech from its noisy version without introducing unnaturally artificial noise. By studying the subband mean-squared error (MSE) of the speech for unsupervised speech enhancement approaches and revealing its relationship with the existing loss function for supervised approaches, this paper derives a generalized loss function that takes residual noise control into account with a supervised approach. Our generalized loss function contains the well-known MSE loss function and many other often-used loss functions as special cases. Compared with traditional loss functions, our generalized loss function is more flexible to make a good trade-off between speech distortion and noise reduction. This is because a group of well-studied noise shaping schemes can be introduced to control residual noise for practical applications. Objective and subjective test results verify the importance of residual noise control for the supervised speech enhancement approach.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Speech and Audio Processing · Advanced Adaptive Filtering Techniques · Acoustic Wave Phenomena Research
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Key Laboratory of Noise and Vibration Research, Institute of Acoustics, Chinese Academy of Sciences, Beijing 100190, China
University of Chinese Academy of Sciences, Beijing 100049, China
Funding
National Science Fund of China
References (40)
Show source-provided references
1. Suppression of acoustic noise in speech using spectral subtraction10.1109/TASSP.1979.1163209 ↗
2. Speech enhancement using a minimum-mean square error short-time spectral amplitude estimator10.1109/TASSP.1984.1164453 ↗
3. Speech enhancement using a minimum mean-square error log-spectral amplitude estimator10.1109/TASSP.1985.1164550 ↗
4. A signal subspace approach for speech enhancement
5. On training targets for supervised speech separation10.1109/TASLP.2014.2352935 ↗
6. An experimental study on speech enhancement based on deep neural networks10.1109/LSP.2013.2291240 ↗
7. Speech enhancement using progressive learning-based convolutional recurrent neural network10.1016/j.apacoust.2020.107347 ↗
8. Zhang, L., Wang, M., Zhang, Q., and Liu, M. (2020). Environmental Attention-Guided Branchy Neural Network for Speech Enhancement. Appl. Sci., 10.10.3390/app10031167 ↗
9. Wu, J., Hua, Y., Yang, S., Qin, H., and Qin, H. (2019). Speech Enhancement Using Generative Adversarial Network by Distilling Knowledge from Statistical Method. Appl. Sci., 9.10.3390/app9163396 ↗
10. A perceptually motivated approach for speech enhancement10.1109/TSA.2003.815936 ↗
11. Reasons why current speech-enhancement algorithms do not improve speech intelligibility and suggested solutions10.1109/TASL.2010.2045180 ↗
12. A deep learning loss function based on the perceptual evaluation of the speech quality10.1109/LSP.2018.2871419 ↗
13. Liu, Q., Wang, W., Jackson, P.J., and Tang, Y. (September, January 28). A perceptually-weighted deep neural network for monaural speech enhancement in various background noise conditions. Proceedings of the European Signal Processing Conference (EUSIPCO), Kos, Greece.
14. Kolbæk, M., Tan, Z.-H., and Jensen, J. (2018, January 15–20). Monaural speech enhancement using deep neural networks by maximizing a short-time objective intelligibility measure. Proceedings of the IEEE International Conference on Acoustics, Speech, Signal Processing (ICASSP), Calgary, AB, Canada.10.1109/ICASSP.2018.8462040 ↗
15. Venkataramani, S., Casebeer, J., and Smaragdis, P. (2017, January 4–9). Adaptive front-ends for end-to-end source separation. Proceedings of the Neural Information Processing Systems (NIPS), Long Beach, CA, USA.
16. Shivakumar, P.G., and Georgiou, P.G. (2016, January 8–12). Perception optimized deep denoising autoencoders for speech enhancement. Proceedings of the Conference of the International Speech Communication Association (INTERSPEECH), San Francisco, CA, USA.10.21437/Interspeech.2016-1284 ↗
17. Rim, A.W., Beerends, J.G., Hollier, M.P., and Hekstra, A.P. (2001, January 7–11). Perceptual evaluation of speech quality (PESQ)-a new method for speech quality assessment of telephone networks and codecs. Proceedings of the International Conference on Acoustics, Speech and Signal Processing (ICASSP), Salt Lake City, UT, USA.
18. Taal, C.H., Hendriks, R.C., Heusdens, R., and Jensen, J. (2010, January 15–19). A short-time objective intelligibility measure for time-frequency weighted noisy speech. Proceedings of the International Conference on Acoustics, Speech and Signal Processing (ICASSP), Dallas, TX, USA.10.1109/ICASSP.2010.5495701 ↗
19. Le Roux, J., Wisdom, S., Erdogan, H., and Hershey, J.R. (2019, January 12–17). SDR–half-baked or well done? In Proceedings of the International Conference on Acoustics, Speech and Signal Processing (ICASSP), Brighton, UK.10.1109/ICASSP.2019.8683855 ↗
20. End-to-end waveform utterance enhancement for direct evaluation metrics optimization by fully convolutional neural networks10.1109/TASLP.2018.2821903 ↗
21. Xu, Z., Elshamy, S., Zhao, Z., and Fingscheidt, T. (2019). Components loss for neural networks in mask-based speech enhancement. arXiv.
22. Zheng, C., Zhou, Y., Hu, X., and Li, X. (2011, January 22–27). Two-channel post-filtering based on adaptive smoothing and noise properties. Proceedings of the International Conference on Acoustics, Speech and Signal Processing (ICASSP), Prague, Czech Republic.10.1109/ICASSP.2011.5946839 ↗
23. A statistical analysis of two-channel post-filter estimators in isotropic noise fields10.1109/TASL.2012.2229981 ↗
24. Subjective evaluation of a noise-reduced training target for deep neural network-based speech enhancement10.1109/TASLP.2018.2882738 ↗
25. Gustafsson, S., Jax, P., and Vary, P. (1998, January 12–15). A novel psychoacoustically motivated audio enhancement algorithm preserving background noise characteristics. Proceedings of the International Conference on Acoustics, Speech and Signal Processing (ICASSP), Seattle, WA, USA.
26. Braun, S., Kowalczyk, K., and Habets, E.A. (2015, January 5–9). Residual noise control using a parametric multichannel wiener filter. Proceedings of the International Conference on Acoustics, Speech and Signal Processing (ICASSP), New Orleans, LA, USA.10.1109/ICASSP.2015.7177991 ↗
27. Benesty, J., Chen, J., Huang, Y., and Cohen, I. (2009). Noise Reduction in Speech Processing, Springer.
28. Noise power spectral density estimation based on optimal smoothing and minimum statistics10.1109/89.928915 ↗
29. Noise spectrum estimation in adverse environments: Improved minima controlled recursive averaging10.1109/TSA.2003.811544 ↗
30. Unbiased mmse-based noise power estimation with low complexity and low tracking delay10.1109/TASL.2011.2180896 ↗
31. Inoue, T., Saruwatari, H., Shikano, K., and Kondo, K. (2011, January 22–27). Theoretical analysis of musical noise in wiener filtering family via higher-order statistics. Proceedings of the International Conference on Acoustics, Speech and Signal Processing (ICASSP), Prague, Czech Republic.10.1109/ICASSP.2011.5947498 ↗
32. Duan, Z., Mysore, G.J., and Smaragdis, P. (2012, January 9–13). Speech enhancement by online non-negative spectrogram decomposition in nonstationary noise environments. Proceedings of the Conference of the International Speech Communication Association (INTERSPEECH), Portland, OR, USA.10.21437/Interspeech.2012-181 ↗
33. Assessment for automatic speech recognition: Ii. noisex-92: A database and an experiment to study the effect of additive noise on speech recognition systems10.1016/0167-6393(93)90095-3 ↗
34. Kingma, D.P., and Ba, J. (2014). Adam: A method for stochastic optimization. arXiv.
35. Kolbæk, M., Tan, Z.-H., Jensen, S.H., and Jensen, J. (2019). On loss functions for supervised monaural time-domain speech enhancement. arXiv.10.1109/TASLP.2020.2968738 ↗
36. Ioffe, S., and Szegedy, C. (2015, January 6–11). Batch normalization: Accelerating deep network training by reducing internal covariate shift. Proceedings of the International Conference on Machine Learning (ICML), Lille, France.
37. Clevert, D.-A., Unterthiner, T., and Hochreiter, S. (2015). Fast and accurate deep network learning by exponential linear units (elus). arXiv.
38. Wichern, G., and Roux, J.L. (2018, January 17–20). Phase reconstruction with learned time-frequency representations for single-channel speech separation. Proceedings of the International Workshop on Acoustic Signal Enhancement (IWAENC), Tokyo, Japan.10.1109/IWAENC.2018.8521243 ↗
39. Performance measurement in blind audio source separation10.1109/TSA.2005.858005 ↗
40. Cepstral smoothing of spectral filter gains for speech enhancement without musical noise10.1109/LSP.2007.906208 ↗