[Standards-JIG] VoIP revisited

Cesar Martinez Izquierdo listas at ono.com
Sat May 14 11:53:25 UTC 2005


El Jueves 14 Abril 2005 12:00, Richard Dobson escribió:
> As has been mentioned there are three main ways to implement voice chat in
> Jabber and they are the following:
>
> 1) The simple approach of just launching an external voip application or
> using an existing protocol library (i.e. H.323, SIP or IAX), either way the
> only thing transmitted over the Jabber connection is the voip initiation
> URI (i.e. callto:), this makes things easy to implement and means that we
> arnt reinventing any wheels.
>
> 2) Use TINS/RTP and implement our own voip standard, this has the benefit
> of being the most tightly integrated and seemless option but it has the
> downside of requiring all Jabber clients to implement this to work, they
> dont really have the option of just launching an external client if they
> dont want to implement the hard bit of VoIP, it also means you need to
> create gateways/transports to be able to communicate with non jabber voip
> users (i.e. SIP).
>
> 3) Have a push to talk style mechanism where you just record wav or mp3 etc
> files and just transfer them using the existing jabber file transfer
> method, then just automatically play them on the receivers side once
> transfer has been completed, this has the benefit of being quite easy to
> implement and doesnt stricktly require any new protocols, but it has the
> downside that there will potensially be quite a delay before the sounds are
> heard by the receiving side and thus is not truely VoIP as it is not
> streaming.
>
> Now none of these protocols are mutually exclusive there is no reason we
> cannot implement multiple protocols in one client.
> Richard

First of all, sorry about taking so much time to reply...

While those protocols are not exclusive, I think is good to define one 
official protocol, otherwise we will end with clients that implement voip but 
are unable to communicate with other clients (which is more or less the 
current situation...).

Regarding the 3) protocol (just record and send mp3, etc), I don't think the 
delay is acceptable in this way. I'm not experience in videoconference 
implementation, but I've read that the subject is far more complicated, and 
you need to take lot of things into consideration to get an acceptable delay.

Best regards,

  César




More information about the Standards mailing list