Journal of Convergence for Information Technology (융합정보논문지)

Convergence Society for SMB (중소기업융합학회)
권호 표지
DOI QR코드

DOI QR Code

DNN based Robust Speech Feature Extraction and Signal Noise Removal Method Using Improved Average Prediction LMS Filter for Speech Recognition

음성 인식을 위한 개선된 평균 예측 LMS 필터를 이용한 DNN 기반의 강인한 음성 특징 추출 및 신호 잡음 제거 기법

  • Oh, SangYeob (Division of Computer Engineering, Gachon University)
  • 오상엽 (가천대학교 컴퓨터공학과)
  • Received : 2021.04.20
  • Accepted : 2021.06.20
  • Published : 2021.06.28
Download PDF
⟨ Previous Next ⟩

Abstract

In the field of speech recognition, as the DNN is applied, the use of speech recognition is increasing, but the amount of calculation for parallel training needs to be larger than that of the conventional GMM, and if the amount of data is small, overfitting occurs. To solve this problem, we propose an efficient method for robust voice feature extraction and voice signal noise removal even when the amount of data is small. Speech feature extraction efficiently extracts speech energy by applying the difference in frame energy for speech and the zero-crossing ratio and level-crossing ratio that are affected by the speech signal. In addition, in order to remove noise, the noise of the speech signal is removed by removing the noise of the speech signal with an average predictive improved LMS filter with little loss of speech information while maintaining the intrinsic characteristics of speech in detection of the speech signal. The improved LMS filter uses a method of processing noise on the input speech signal by adjusting the active parameter threshold for the input signal. As a result of comparing the method proposed in this paper with the conventional frame energy method, it was confirmed that the error rate at the start point of speech is 7% and the error rate at the end point is improved by 11%.

음성 인식 분야에서 DNN이 적용됨에 따라 음성 인식의 이용이 증대되고 있으나 기존의 GMM 보다 병렬 훈련에 대한 계산의 양이 많아야 되며, 데이터의 양이 적으면 오버피팅이 발생한다. 이를 해결하기 위해 데이터의 양이 작은 경우에도 강인한 음성 특징 추출과 음성 신호 잡음 제거에 효율적인 방안을 제시한다. 음성 특징 추출은 음성에 대한 프레임 에너지의 차이와 음성 신호에 영향을 받는 영 교차율과 레벨 교차율을 적용하여 음성 에너지의 효율적 추출을 한다. 또한, 잡음 제거를 위해 음성 신호에 대한 검출에서 음성의 고유 특성을 유지하면서 음성 정보 손상이 적은 평균 예측 LMS 필터를 개선하여 음성 신호의 잡음을 제거하여 데이터양이 적은 경우의 문제를 해결한다. 개선된 LMS 필터는 입력 신호에 대한 활성 파라미터 임계치를 조정하여 입력된 음성 신호에 대한 잡음을 처리하는 방법을 사용한다. 본 논문에서 제안한 방법을 사용하여 기존의 프레임 에너지를 이용한 방법과 비교한 결과 음성의 시작점의 오차율은 7%, 끝나는 점 오차율에서 11% 향상된 성능을 확인하였다.

Keywords

Acknowledgement

This research was supported by Global Infrastructure Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Science and ICT(NRF-2018K1A3A1A20026485)

References

  1. S. Y. Oh. (2020). Speech Recognition Performance Improvement using a convergence of GMM Phoneme Unit parameter and Vocabulary Clustering. Journal of Convergence for Information Technology, 10(8), 35-39. DOI : 10.22156/CS4SMB.2020.10.08.035
  2. C. S. Ahn & S. Y. Oh. (2012). Gaussian Model Optimization using Configuration Thread Control In CHMM Vocabulary Recognition. The Journal of Digital Policy and Management. 10(7), 167-172. DOI : 10.14400/JDPM.2012.10.7.167
  3. J. Homer & I. Mareels. (2004). LS detection guided NLMS estimation of sparse system. Proceedings of the IEEE 2004 International Conference on Acoustic. Speech, and Signal Processing(ICASSP). Montreal, Quebec, Canada. DOI : 10.1109/ICASSP.2004.1326394
  4. B. Sisman, J. Yamagishi, S. King & H. Li. (2020). An overview of voice conversion and its challenges: From statistical modeling to deep learning. IEEE/ACM Transactions on Audio, Speech, and Language Processing.
  5. B. F. Wu & K. C. Wang. (2005). Robust endpoint detection algorithm based on the adaptive band-partitioning spectral entropy in adverse environments. IEEE Transactions on Speech and Audio Processing, 13(5), 762-775. DOI : 10.1109/TSA.2005.851909
  6. Q. Li, J. Zheng, A.Tsai & Q. Zhou. (2002). Robust endpoint detection and energy normalization for real-time speech and speaker recognition. IEEE Transactions on Speech and Audio Processing, 10(3), 146-157. DOI : 10.1109/TSA.2002.1001979
  7. A. Arango, J. P'erez & B. Poblete. (2019). Hate Speech Detection is Not as Easy as You May Think, A Closer Look at Model Validation. In Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval, 45-54. Paris, France: Association for Computing Machinery. DOI : 10.1145/3331184.3331262
  8. S. S. Aluru, B. Mathew, P. Saha & A. Mukherjee. (2020). Deep Learning Models for Multilingual Hate Speech Detection, arXiv preprint arXiv:2004.06465
  9. E. T. S. I. Standard. (2003). Speech Processing, Transmission and Quality aspects(STQ); Distributed speech recognition; Advanced front-end feature extraction algorithm; Compression algorithms. ETSI ES 202 050 v.1.1.3.
  10. P. Scart & J. Filho, (2002). Speech enhancement based on a priori signal to noise estimation. In 1996 IEEE International Conference on Acoustics, Speech, and Signal Processing Conference Proceedings (Vol. 2, pp. 629-632). IEEE.
  11. K. Chung & S. Y. Oh. (2015). Improvement of speech signal extraction method using detection filter of energy spectrum entropy. Cluster Computing, 18(2), 629-635. DOI : 10.1007/s10586-015-0429-9
  12. S. Kamarth & P.Loizou. (2002). A multi-band spectral subtraction method for enhancing speech corrupted by colored noise. In ICASSP (Vol. 4, pp. 44164-44164).
  13. Yi Hu & P. C. Loizou. (2008). Evaluation of objective quality measures for speech enhancement. IEEE Transactions on audio, speech, and language processing, 16(1), 229-238. https://doi.org/10.1109/TASL.2007.911054
  14. S. Y. Oh & K. Chung. (2018). Performance evaluation of silence-feature normalization model using cepstrum features of noise signals. Wireless Personal Communications, 98(4), 3287-3297. DOI : 10.1109/TASL.2007.911054
  15. K. C. Wang & Y. H. Tsai. (2008). Voice activity detection algorithm with low signal-to-noise ratios based on spectrum entropy. In 2008 Second International Symposium on Universal Communication (pp. 423-428). DOI : 10.1109/ISUC.2008.55