Re: [Ant-phone-devel] Latency, fragments and their sizes.

[Bruno Hertz]

> This is right for the optimum case. People often say it's crazy to try
> to run differently clocked devices in sync. But theoretically it's
> possible e.g. with cheating in (or out) an additional sample from time
> to time for destination devices running slightly to fast (or slow).

OK, seems to be a pretty tough issue after all :)

In the meantime, I did some googling and found that quite some
FritzFon users have the same problem, i.e. echoing. So it seems
software telephony in general, or at least with separate isdn and
sound cards, suffers from that deficiency due to buffering and
sync issues, as you said.

Still, what I'll try and likely could improve matters is:

(1) decrease capi buffer size

  In capidrv, there's a default size of 2kb. I already tried 512 bytes
  and have a feeling that there's less latency but maybe more 'clicks'.
  256 didn't work, the other party gets no sound from me (maybe due to
  ant buffer sizes?)

(2) have a look at ant's and sound buffering

  Maybe decreasing buffer sizes helps here, too ?

(3) get some 2.4.x preemptible patched kernel

  I wonder wether this works with the AVM drivers, but I guess a shot
  at it shouldn't do any harm.

Hopefully, those measures together already let things run more
smoothly. I'll check back when I have some results.

Thanks so far :)
Bruno