On Wed, Mar 25, 2009 at 5:38 PM, Stefan Sayer <[email protected]> wrote: > Hello, > > o Peter Loeppky [03/25/09 22:04]: >> >> I am having an issue with out of band dtmf. My problem is that if the >> caller is entering dtmf information and say he enters 55, and it is >> registered as one digit. >> >> I have tracked it down to AmDtmfDetector::registerKeyReleased or >> AmDtmfDetector::checkTimeout. They are not processing Events the End of >> Event flag in the rtp packet. >> >> The AmRtpDtmfDetector is handling the flag, but once the event get >> relayed to AmDtmfDetector class it can delay it long enough that the >> second 5 gets treated as the same keypress as the first one. >> >> Also it will have to handle multiple end events, as my source sends it >> three times. > > that is the reason it waits for some time before it passes the DTMF as key > press to the session. It may indeed be not possible to tell with 100% > certainity whether the RTP events belong to the same key press; there is no > ID or the like in the events.
The timestamp for two different events should be different and this can act as an event ID. This should be an indication that a packet belongs to a new event (even if the event is identical to the previous one) and it should reliably detect a new event. Regards, Ovidiu Sas _______________________________________________ Semsdev mailing list [email protected] http://lists.iptel.org/mailman/listinfo/semsdev
