Digital Signal Processing Reference
In-Depth Information
2
0.4
h
92
0.35
0.3
1
h
85
0.25
0.2
0
h
20
0.15
0.1
h
35
−
1
0.05
h
0
0
−
2
−
0.05
1
×
10
5
0
20000
40000
60000
80000
0
20000
40000
60000
80000
Time (samples)
Time (samples)
Figure 11.37
Performance of the noise-echo canceller at 0 dB SNR
the adaptive filtering used in the enhancement algorithms. It is crucial that
the filters converge rapidly and do not diverge under any circumstances
(any level of acoustic noise and echo). Normalized LMS algorithms usually
provide adequate performance. The newly-proposed adaptive normalized
LMS algorithm [24], discussed in this chapter, has shown robust performance
under significant levels of acoustic noise and echo.
Bibliography
[1] J. S. Lim and A. V. Oppenheim (1979) 'Enhancement and bandwidth
compression of noisy speech', in
Proc. IEEE
, 67(12):1568-1604.
[2] Y. Ephraim (1992) 'Statistical-model-based speech enhancement sys-
tems', in
Proc. IEEE
, 80(10):1526-55.
[3] J. R. Deller, H. G. Proakis, and J. H. L. Hansen, (1993) 'Speech enhance-
ment', in
Discrete-Time Processing of Speech Signals
,Chapter8.NewYork:
Macmillan
[4] S. V. Vaseghi (2000)
Advanced digital signal processing and noise reduction
,
2nd edition. Chichester: John Wiley & Sons Ltd
[5] Y. Ephraim and H. L. Van Trees (1995) 'A signal subspace approach
for speech enhancement', in
IEEE Trans. Speech and Audio Processing
,
3(4):251-66.
Search WWH ::
Custom Search