It seems to me that if the developers would go as far as to put the
functionality to reassemble rtp packets to audio, that they have
already stretched the definition of a packet sniffer. Why not go the
extra step and make the feature more robust? In its current form, I
would say its broken, if the simple act of pushing a "1" during the
conversation, breaks the ability to save the file. (without filtering
out the DTMF packets). Just my two cents.
On Thu, 17 Mar 2005 14:45:33 -0500, Scott Lowrey <slowrey@xxxxxxxxxxx> wrote:
> Cool idea, but what this amounts to is almost equivalent to implementing a
> proper RTP layer within Ethereal. To "insert silence" means analyzing the
> special event (DTMF), and *generating* the silence. This is not trivial nor
> should this functionality be provided by a utility like Ethereal.
>
> I'm not an Ethereal expert and there may be a way to get this to work, but
> it seems like we're stretching the definition of a packet sniffer a bit.
> The thing is that RTP can be quite complex - it's rarely a simple bitstream
> that can be assembled into a .wav file.
>
> Still, Ethereal continues to amaze me so I don't want to dampen enthusiasm.
> :)
>
> Maybe a better idea would be to create a small app that combines libpcap
> and GNU ccrtp...?
>
> Erik Naslund wrote:
>
> That's right. You shouldnt remove anything from the stream when you're
> interested in reconstructing it later (rtpdump/rtpplay). However, the "save
> payload" functionality should be modified in order to ignore anything that's
> not voice (voice=only G.711 for the moment) and that shouldn't disturb the
> audio content too much, does it? I think that's what he's talking about. I
> had no time to look at the capture file for the moment however.
> The audio
> content gets disturbed when you remove the DTMF packets. The more packets of
> this type, the more the timing of the conversation is off. Think about
> calling your credit card company, when you enter in your full CC #, plus
> following the menu prompts. This could cause the timing of the conversation
> to be off by several seconds. Again, if the Save Payload functionality could
> simply insert silence or false tones during the audio encoding process, this
> would preserve the timing of the conversation. IANAD but it would seem
> doable, as the DTMF packets seem very detailed as to how long the button is
> pressed, which button it is, and when it "ends". I am a very grateful user
> of a fabulous FOSS package. I dont want to come off as disgruntled or
> ignorant. Thanks for all the work everyone does on making ethereal such an
> awesome tool for sysadmins such as myself, as well as the casual home user.
> You guys are great. -Erik _______________________________________________
> Ethereal-users mailing list Ethereal-users@xxxxxxxxxxxx
> http://www.ethereal.com/mailman/listinfo/ethereal-users
>
> --
> Scott Lowrey
> Test Engineering Manager
> NexTone Communications
> Gaithersburg, Maryland USA
>
> 1.240.912.1369
>
