[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Linphone-developers] Bursty rtp traffic

From: Simon Morlat
Subject: Re: [Linphone-developers] Bursty rtp traffic
Date: Tue, 16 Jun 2009 11:53:44 +0200
User-agent: KMail/1.11.4 (Linux/2.6.29-2-amd64; KDE/4.2.4; x86_64; ; )

Hi Vadim,

Unfortunately despite I wrote the original speex-ec filter, I could not follow 
the evolutions that were made to it by Aymeric, if I remember to try to solve 
the synchronisation nightmare between the two inputs.
I actually do not understand clearly what Aymeric intends to do and thus I 
have no idea whether your modifications fixes some real problems.
Maybe you should ask him directly.

My opinion about this echo canceler is that it works very well when these 
conditions are met:
- low latency sound card
- delivery of samples chunks at regular, small intervals.
- in case of several soundcards are involved, they must have an accurate clock 
(close to 8000 or 16000hz) to be "nearly" synchronous.
This is mostly always the case on linux, due to the overall good quality of 
alsa drivers and linux kernel 's real time capabilites.

For bad conditions, it is not clear that a real good solution exists.


Le jeudi 11 juin 2009 03:16:40, Vadim Lebedev a écrit :
> Hi while still tracking the bursty rtp output issue,
> I'm trying to analyze how speexec.c  works -- it is pretty hairy stuff,
> so i modified it a little bit to simplify the reading:
> I introduced couple of #defines
> #define REFSIG  f->inputs[0]
> #define ECOSIG  f->inputs[1]
> #define REFBUF  s->in[0]
> #define ECOBUF  s->in[1]
> #define TOSPKR f->outputs[0]
> #define TOCODR f->outputs[1]
> this improves readability a lot
> I've added some comments
> and fixed couple of trivial bugs (IMO)  all bug are marked // Fix: Vadim
> I've also adde couple of questions in the comments where i suspect a
> problem
> but not sure about it...
> Simon, can you please review the code?
>   Thanks
> Vadim

reply via email to

[Prev in Thread] Current Thread [Next in Thread]