Speech Enhancement for ASR

1. Speech Enhancement for ASR by Hans Hwang 8/23/2000 Reference 1. Alan V. Oppenheim ,etc.,”Multi-Channel Signal Separation by Decorrelation”,IEEE Trans. on ASSP,405-413,1993 2.Yunxin Zhao,etc.,”Adaptive Co-channel Speech Separation and Recognition”,IEEE Trans. On SAP,138-151,1999 3.Ing Yang Soon,etc.,”Noisy Speech Enhancement Using Discrete Cosine Transform”,Speech communication,249-257,1998

2. Outline • Signal Separation by S-ADF/LMS • Speech Enhancement by DCT • Residual Signal Reduction • Experimental Results

3. Speech Signal Separation • Introduction: -To Recover the desired signal and identify the unknown system from the observation signal -Speech signal recovered from SSS will increase SNR and improve the speech recognition accuracy -Specifically consider the two-channel case

4. Two-channel model description A and Bare cross-coupling effect between channels and we ignore the transfer function of each channel. xi(t) is source signal and yi(t) is acquired signal SSS cont’d

5. SSS (cont’d) • Source separation system (separate source signals out from acquired signals) and called decoupling filters and modeled as FIR filter

6. SSS by ADF • Calculate the FIR coeff. by adaptive decorre- lation filter(ADF) proposed by A. V. Oppenheim in 1993 -The objective is to design decoupling filter s.t., the estimated signals are uncorrelated. -The decoupling filtering coeff.’s are estimated iteratively based on the previous estimated filter coeff.’s and current observations

7. SSS by ADF (cont’d) • The closed form of decoupling filters where

8. SSS by ADF (cont’d) • Choice of adaptation gain -As time goes to infinite the adaptation gain goes to zero for the system stable consideration. -Optimal choice adaptation gain for the system stability and convergence. -

9. SSS by ADF (cont’d) • The experiment of :

10. Source Signal Detection(SSD) • Introduction -If one of the two is inactive then the estimated signals will be poor by ADF and cause the recog- nition errors. -So the ASR and ADF are performed within active region of each target signal.

11. SSD (cont’d)

12. SSD (cont’d) • SSD by coherence function If then If then

13. SSD (cont’d) - decision variable -Decision Rule:

14. SSD (cont’d) -Implementation using DFT and Result

15. SSD (cont’d)

16. Improved Filter Estimation • Widrow’s LMS algorithm proposed in 1975 -If we don’t know A or B in observation(i.e., one of the source signals is inactive) then the estimation of filters will cause much errors compared to the actual filters. -If we know source signal 2 is inactive(using SSD) then we only estimate filter B and remain filter A unchanged.

17. Improved Filter Estimation • LMS algorithm and result

18. Experimental Results -Evaluate in terms of WRA and SIR

19. Experimental Result (cont’d) *Use 717 TIMIT sentences to train 62 phone units. Front-end feature is PLP and its dynamic. Grammar perplexity is 105. After acoustic normalization

20. Speech Enhancement usingDiscrete Cosine Transform • Motivation -DCT provides significantly higher compaction as compared to the DFT

21. SE Using DCT (cont’d) -DCT provides higher spectral resolution than DFT -DCT is real transform so it has only binary phases. Its phase won’t be changed unless added noise is strong.

22. Estimating signal by MMSE • Intorduction -y(t)=x(t)+n(t) and Y(k)=X(k)+N(k) Assume DCT coeff.’s are statistically independent and estimated signal is less diffenent from the original signal. - , by Bayes’ rule and signal model

23. MMSE (cont’d) • Estimating signal source by Decision Directed Estimation(DDE) (proposed by Ephraim & Malah in ‘84) = 0.98 in computer simulation

24. Reduction of Residual Signal • Introduction -If the source signal more likely exists then the estimated is more reliable. -two states of inputs H0:speech absent H1:speech present : modified filter output

25. Reduction of Residual Signal - where

26. Experimental Results Measure in Segmental SNR White noise added Fan noise added

27. Experimental Results