On 02/02/2011 09:48 PM, João Mesquita wrote:
Mitch, you beat me to it...
I think it might be time for me to get back to FSComm and drop being picky about the AEC thing? Sad but true, I think that the only way for me to draw attention of someone who knows about this stuff is actually doing the rest like you did. Great initiative. I truly congratulate you from pulling something I couldn't so far.
I am puzzled why the AEC issue would ever have affected your development of FSComm. Although a headset or handset needs EC for really good telephone, the lack fo AEC is only a killer for speakerphone usage. On this topic, a little experimentation suggests the following... Skype appears to do good AEC on a PC, but when you listen carefully is seems more like they avoid proper AEC, and use the kind of adaptive gain juggling the DSP Group pushed heavily in the early 90s. The final result is very acceptable, though, and avoids the problems to non-matching mic and speaker sampling rates that you get with many PC sound cards. Skype on a Symbian handset and a Mac seems to do proper echo cancellation, and gives results a bit better than on a PC. The limited range of hardware used in these platforms makes AEC more of a practical option. If anyone has further information about the exact schemes used by Skype, I would really like to hear from them. Skype is used on so many machines, that whatever strategy they have found workable across a wide range of hardware should be a safe bet for us to use. Steve