On Fri, 2006-09-29 at 12:02 -0500, Paul Li wrote: > Thank you for your reply. Since we (our UA) can hear the sound of > key-pressing, could we re-construct these events from the sound even though > the intermediate VoIP service provider or PSTN Gateway does not pass them > along? Do we have any other alternative to this problem? We might not have > any execuse if we offer our customers IVR function.
In principle, the receiving UA can test and decode the received audio stream to detect DTMF. But at least for some popular codecs, the audio transmission qualities are poor enough that it is difficult to reliably detect DTMF. The detection of DTMF is also disrupted by missing or delayed audio packets, as the DTMF decoder cannot reliably distinguish between two occurrences of a single digit and one occurrence with a missing audio sample in the middle. The RFC 2833 encoding is designed to be robust against missing packets. In practice, we've discovered, there is no viable alternative to requiring RFC 2833 compliance on the part of every component that handles RTP. Dale _______________________________________________ sipxtapi-dev mailing list [email protected] List Archive: http://list.sipfoundry.org/archive/sipxtapi-dev/
