A.3 VAD performance
3GPP46.032Full rate speechRelease 17TSVoice Activity Detector (VAD) for full rate speech traffic channels
In optimizing a VAD a difficult trade‑off has to be made between speech clipping which reduces the subjective performance of the system, and the average activity factor. The benefit of DTX is increased as the average activity factor is reduced. However, in general, a reduction of the activity will be associated with a greater risk for audible speech clipping.
In the optimization process, great emphasis has been placed on avoiding unnecessary speech clipping. However, it has been found that a VAD with virtually no audible clipping would result in a very high activity and very little DTX advantage.
The VAD specified in this technical specification introduces audible and possibly objectionable clipping in certain cases, mainly with low input levels. However, a comprehensive evaluation programme consisting of about 600 individual conversations conducted in a wide range of realistic conditions, it was found that about 90% of the conversations were free from objectionable clipping.
The voice activity performance of the VAD is summarized in table A.3.1. The activity figures are averages of a large number of conversations covering factors like different talkers, noise characteristics and locations. It should be noted that the actual activity of a particular talker in a specific conversation may vary considerably relative to the averages given. This is due both to the variation in talker behaviour as well as to the level dependency of the VAD (the channel activity has been found to decrease by about 0,5 points of percentage per dB level reduction). However, as mentioned above, a decreased speech input level increases the risk of objectionable speech clipping.
All the values given are activity figures, i.e. the % of time the radio channel has to be on.
Table A.3.1: Summary of channel activity
|
Telephone |
Situation |
Typical channel |
|
Handset |
Quiet location |
55% |
|
Handset |
Moderate office |
60% |
|
Handset |
Strong voice |
65‑70% |
|
Handsfree/ |
Variable vehicle |
60% |