Speech Enhancement - Digital Speech: Coding for Low Bit Rate Communication Systems - page 373

11
Speech Enhancement
11.1 Introduction
In voice communications, speech signals can be contaminated by environmen-
tal noise and, as a result, the communication quality can be affected making
the speech less intelligible. Furthermore, compression of the noisy speech
with a low bit-rate vocoder may result in considerable quality degradation
due to frequent estimation errors of speech production model parameters
required by the vocoder. This problem can be reduced significantly by speech
enhancement (or noise cancellation), which may enable more pleasant voice
communication by suppressing the noise components in input signals.
Generally, it is assumed that the noisy speech signal is formed additively
by speech and noise signals in which the noise is generated by environ-
mental sources such as vehicles, street noise, babble, etc. Therefore, in real
environments, complete noise cancellation is not feasible as it is not possible
to completely track varying noise types and characteristics that change with
time. However, by assuming that the noise characteristics change slowly in
comparison with speech, it is possible to achieve significant reduction in
the background noise levels producing more pleasant and intelligible speech
quality. Speech enhancement techniques can help the speech model param-
eter extraction process used in low bit-rate vocoders and hence they are
becoming an integral part of low bit-rate speech coding systems.
Speech enhancement techniques can be classified, depending on the number
of available microphones, into single and multiple channels. In the case of
a single channel, the reference noise is not available explicitly. The noise
statistics are typically characterized during voice-inactive regions between
talk spurts using a voice activity detector. On the other hand, when dual
channels are available, one microphone senses the noisy speech, but the
other can be used mainly to catch the noise. By eliminating the noise factor
collected by the second microphone from the first, it would be possible to