Re: [Ccrtp-devel] problem while reducing the seding rate

[David Sugar]

The way I usually track RFC2833 packets is by intercepting them prior to 
queue, handling them as they arrive and dropping them there, by 
overriding onRTPPacketRecv.  I also do soft processing for dropping 
"silent" packets when the gateway is unable to do this for me as well.

bool RTPStream::onRTPPacketRecv(IncomingRTPPkt &pkt)
{
        Event event;
        struct dtmf2833 *evt;
        unsigned pid = Driver::sip.getDTMFPayload();
        AudioCodec *codec;
        Level level;

        // this is where we test for 2833!!!

        if(!pid || pid != pkt.getPayloadType())
                goto check2;

        evt = (struct dtmf2833 *)pkt.getPayload();

        if(evt->event != lastevt)
                goto post;

        if(tonecount)
                goto checkbit;

        if(evt->ebit)
                goto checkbit;

post:
        lastevt = evt->event;

        memset(&event, 0, sizeof(event));
        switch(evt->event)
        {
        case 0:
        case 1:
        case 2:
        case 3:
        case 4:
        case 5:
        case 6:
        case 7:
        case 8:
        case 9:
        case 10:
        case 11:
        case 12:
       case 13:
        case 14:
        case 15:
                event.id = DTMF_KEYUP;
                event.dtmf.digit = evt->event;
                event.dtmf.duration = 60;
                break;
        case 16:
                event.id = LINE_PICKUP;
                break;
        default:
                return false;
        }
        if(session->postEvent(&event))
        {
                setSource(NULL);
                setSink(NULL);
                setTone(NULL);
        }

checkbit:
        if(evt->ebit && tonecount && stopid != MSGPORT_WAKEUP)
        {
                memset(&event, 0, sizeof(event));
                event.id = stopid;
                stopid = MSGPORT_WAKEUP;
                session->postEvent(&event);
                tonecount = 0;
        }
        else if(evt->ebit)
                tonecount = 0;
        else
                tonecount = 3;
        return false;

check2:
        lastevt = 256;

        if(pkt.getPayloadType() != Driver::sip.getDataPayload())
                return false;

        if(sink)
                goto silence;

        if(Driver::sip.getDTMFInband())
                goto silence;

        return false;

silence:
        level = Driver::sip.silence;
        codec = session->codec;
        if(!level || !codec)
                return true;

	if(codec->isSilent(level, (void *)pkt.getRawPacket(), 
Driver::sip.info.framecount))
                return false;

        return true;
}




Michel de Boer wrote:
> Hi Dinil,
>
> This sounds like the problem that I had with sending RTP DTMF events.
> All RTP packets in this case have the same timestamp and ccRTP
> discards packets when the timestamp gets too old.
>
> In your case you increase the PERIOD, but it seems that
> you keep the payload size of your packets the same. That means
> that the timestamps you provide with putData will not be in
> sync with the timestamp clock from ccRTP itself. The timestamp
> of your second packet indicates that this packet should come 50ms after
> the first packet. But you only deliver it 100ms after the first
> packet.
>
> I think you have to deliver the data to ccRTP with the correct
> rate (depending on the number of samples per second). So if you
> want to double the period from 50 to 100 then you should also double
> the size of your payload.
>
> Regards,
> Michel
>
>
>
> Dinil Divakaran wrote:
>
> > Hi all,
> >
> > I was working on an example for transmitting an audio file using ccrtp 
> > (http://www.gnu.org/software/ccrtp/doc/refman/html/audiotx_8cpp-example.html) 
> >
> >
> > The following lines are from the above example. In this code, packets are
> > sent and then thread sleeps for a time of 'PERIOD' milliseconds. I 
> > changed
> > the value of PERIOD to 100 (milliseconds) and found that even though the
> > program says that the packets are sent; the packets weren't actually 
> > sent.
> > 'tcpdump' showed that only the first packet was sent; and thereafter only
> > RTCP packets were sent (with still lesser frequency). When I changed the
> > value of 'PERIOD' to 50, the packets were sent successfully.
> >
> > Can anyone please tell me where have I gone wrong ?
> >
> > -------------------------------------------------------------------------- 
> >
> >                               .
> >                               .
> >                               .
> >
> >         cout << "Transmitting " << PACKET_SIZE
> >              << " octects long packets "
> >              << "every " << PERIOD << " milliseconds..." << endl;
> >
> >         unsigned char buffer[PACKET_SIZE];
> >         int count=PACKET_SIZE;
> >
> >         // This will be useful for periodic execution
> >         TimerPort::setTimer(PERIOD);
> >
> >         setCancel(cancelImmediate);
> >         // This is the main loop, where packets are transmitted.
> >         for( int i = 0 ; (!sendingfile || count > 0) ; i++ ){
> >
> >             count = ::read(audioinput,buffer,PACKET_SIZE);
> >             if( count > 0 ){
> >                 // send an RTP packet, providing timestamp,
> >                 // payload type and payload.
> >                 socket->putData(PACKET_SIZE * i, buffer, PACKET_SIZE);
> >             }
> >             cout << "." << flush;
> >
> >             // Let's wait for the next cycle
> >             Thread::sleep(TimerPort::getTimer());
> >             TimerPort::incTimer(PERIOD);
> >         }
> >         cout << endl << "I have got no more data to send. " <<endl;
> >
> >                               .
> >                               .
> >                               .
> >                               .
> > -------------------------------------------------------------------------- 
> >
> >
> > - Dinil
> >
> >
> >
> > _______________________________________________
> > Ccrtp-devel mailing list
> > address@hidden
> > http://lists.gnu.org/mailman/listinfo/ccrtp-devel
>
>
> _______________________________________________
> Ccrtp-devel mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/ccrtp-devel

dyfet.vcf
Description: Vcard