about VAD function

HI Conhen:

The VAD (Voice Activity Detector) was designed for teleconferencing systems where the background noise is low and not music or for cases where Voice Is the most significant and dominant source. This block is not meant to distinguish between music and voice. That would be a very complicated problem to solve.

Voice does have a spectrum signature and therefore by adding some filtering you can approximate voice from other sources but this simple techniques would yield a large amount of false triggers.

Hope this helps.

Miguel

Miguel, thanks for your helpful answer!

Cohnen, you'll find an introductory explanation of the challenges affecting VAD in the Wikipedia article,

http://en.wikipedia.org/wiki/Voice_activity_detection

I've also seen some academic papers on this subject you could find with an Internet search.

The best VAD methods appear to be based upon frequency-domain analysis (starting with a frame-by-frame FFT on the input signal). Of the present SigmaDSPs, only the ADAU145x series has FFT capability. However, there's no VAD block in its toolbox yet. I came up with a homemade VAD for the ADAU1701 (for which there's never been a ready-made VAD block), but I doubt it would work any better than ADI's VAD block -- if you're curious, see:

VAD for ADAU170x

Best regards,

Bob

Thanks Bob

Thanks Miguel

This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

Thank you,
EZ Admin