Nice to hear the kernel mixer has been completely rewritten. It was a real headache trying to run a voip application against it.

As a voice-engineer responsible for the voice-engine within popular voip products like voipbuster and other Finerea related voip products I would like to know the following:

What API should I use to finally match the microphone and speakersignals correctly, having the lowest latency and minimized drift of those signals?

As for now it's a real "pain in the ..." to match speakersignals with microphone signals. The way we did it with the wavein/out api and a lot of "measurement" code around it, does not work under vista. It seems the DMA size used has been increased instead of lowered???

Please help me out on taking the right direction as not to waste a lot of time trying to figure out which way to go. For this type of pc-based applications you must have a certain API or framework in mind which we should use. We think we have one of the best Echo cancelation voice-engines with an unbeatable double talk performance in desktop speaker/microphone environments in the market today and would like to keep it that way!

Regards,


Martien.