15 Mar
2011
15 Mar
'11
2:19 p.m.
On 03/15/2011 10:05 PM, Avi Marcus wrote:
MAYBE if you would *strictly* have only one speaker at a time, with no volume adjustments etc, then maybe you can hack something together about a conference in g729 w/o transcoding. Maybe. You can extract the energy from the encoded frames, but knowing which energy to choose is another matter. Just taking the greatest energy will lead to rapid hoping amongst speakers. If you are more relaxed about switching speaker you will miss the start of voice spurts. The loudest might not even be the most relevant speaker.
Steve