[Speex-dev] Jitter buffer latency
Jean-Marc Valin
jean-marc.valin at usherbrooke.ca
Mon Jan 14 03:05:01 PST 2008
The "worst case" as you call it is just your network jitter plus
probably one frame.
Jean-Marc
Andy Ngo a écrit :
> Hi Jean-Marc,
>
> Thanks for your response. Given a worst case scenario, what is the "worst case" latency (in terms of Speex frames) that the jitter buffer algorithm will incur? We're trying to determine the worst case hard number. Sorry for unclear question below; what I was trying to ask is that given a worst case latency (which I'm asking in the first question) inherent in using the jitter buffer feature, is that any way in the code to reduce the latency? Thanks.
>
> Regards,
> Andy
>
> ----- Original Message ----
> From: Jean-Marc Valin <jean-marc.valin at usherbrooke.ca>
> To: Andy Ngo <ndno72-speex at yahoo.com>
> Cc: speex-dev at xiph.org
> Sent: Sunday, January 13, 2008 1:00:50 AM
> Subject: Re: [Speex-dev] Jitter buffer latency
>
>
> The latency changes. It's set to the minimum value that will ensure
> packets arrive in time. Not sure what you mean by "is there a way to
> strink the "late" and "early" bin number to keep the latency in
> check?".
>
> Jean-Marc
>
> Andy Ngo a écrit :
>> Hi,
>>
>> Our project is using the jitter buffer feature built in Speex. We
>> noticed there are some latency when using the jitter buffer. Does
>> anyone know what is the "worst case" latency inherent in the jitter
>> buffer algorithm? I believe someone already mentioned that it's
>> adaptive but is there a worst case hard number (in terms of 20ms
>> Speex frames)? I'm not familiar with the jitter buffer code (there
>> doesn't seem to be any white paper on the jitter buffer feature); is
>> there a way to strink the "late" and "early" bin number to keep the
>> latency in check? Thanks in advance.
>>
>> Regards, Andy
>>
>>
>>
>>
>>
> ------------------------------------------------------------------------
>>
>> _______________________________________________ Speex-dev mailing
>> list Speex-dev at xiph.org
>> http://lists.xiph.org/mailman/listinfo/speex-dev
>
>
>
>
More information about the Speex-dev
mailing list