Voice Activity Detection - Digital Speech: Coding for Low Bit Rate Communication Systems - page 353

Inactive Active Inactive Active Inactive

Clean speech

Noisy speech
(vehicle, 5 dB SNR)

Figure 10.1 Voice active and inactive regions

Voice
Decoder

Voice
Encoder

Input
speech

Output
s peech

Active

Channel

Inactive

CNI

CNG

VAD

Active / Inactive

Figure 10.2 Overall structure of a speech coding system with silence compression

• Improvement of the soft channel capacity in the code division multiple

access (CDMA) system: The theoretical capacity of a CDMA system is usu-
ally defined by the possible combinations of the spreading code. However,
due to interference from other users, the CDMA capacity is limited to a
value smaller than this theoretical limit, i.e. due to interference from other
users, the error rates received by some users may be too high to enable
accurate decoding. By reducing the transmission power during speech
pauses, the interference on air can be reduced, which may automatically
allow more users on the system, hence achieving an increase in the CDMA
system's capacity.

• Power-saving for mobile terminals: Mobile terminals do not have to trans-

mit radio signals during pauses. Thus, the battery life time of the terminals
can be extended by conserving power during speech inactive periods.