On 2/29/08, pierp.petri at katamail.com <pierp.petri at katamail.com> wrote: > HI, > > I am experiencing a strange problem on windows mobile 6 (HTC TyTn). Using > "streamutil.c " on mobile device with sampling rate 8 KHz and the default > configuration(PCMU codec), when I send voice signal to an RTP server the > result is good . When I send the signal to a automatic speech recognition > (loquendo) the log dump is an audio sound that seems accelerated > (oversampled) . In other words, if I slow down my speech, the recognizer is > able to "understand" what I am saying. Speaking normally, it does not. Does > anyone knows what can be the cause. is it possible an errata RTP > configuration? Somehow I doubt that this is an RTP problem, since you said it's fine when the other end is an RTP server. How much approximately the acceleration is? If it's small (say, 1%), probably it's the sound clock running faster on either end. If it's big, then there's a problem somewhere, on either end. Perhaps you can capture the RTP packets with Wireshark, save the payload to a file and try to play back this to see if there's anything suspicious. cheers, -benny > Many thanks > > Pierpaolo
