A real time implementation of a system for enhancement of speech in additive acoustic noise is described. The technique used is to perform a spectral decomposition of noisy speech via channel vocoder filter bank analysis and to attenuate each spectral component depending on how much the measured speech plus noise power exceeds an estimate of the background noise. A two state model for the speech event (speech absent or speech present) is applied in determining the maximum likelihood estimator of the speech power. This model has resulted in a new class of suppression curves which permits a tradeoff of noise suppression against speech distortion. Experiments utilizing the real time implementation have shown that the noise can be made imperceptible by proper choice of the suppression factor. Integration of the noise suppression filter into the analysis section of a narrowband vocoder is described. This combined system represents an integrated robust vocoder structure where acoustic noise can be suppressed prior to pitch estimation and determination of the modulator gains.
[1]
Robert D. Preuss.
A frequency domain noise cancelling preprocessor for narrowband speech communications systems
,
1979,
ICASSP.
[2]
B. Gold,et al.
New applications of channel vocoders
,
1981
.
[3]
T. Bially,et al.
A Digital Channel Vocoder
,
1970
.
[4]
Marilyn L. Malpass,et al.
Speech enhancement using a soft‐decision maximum likelihood noise suppression filter
,
1979
.
[5]
Russell J. Niederjohn,et al.
An investigation of several frequency-domain processing methods for enhancing the intelligibility of speech in wideband random noise
,
1978,
ICASSP.
[6]
Richard M. Schwartz,et al.
Enhancement of speech corrupted by acoustic noise
,
1979,
ICASSP.