Speex dev - Jun 2009 - Delays estimation in Speex algorithms

Hi,

I want to do a careful measure of mic-to-net and net-to-spkr delays
in sip media engine and to make it really precise I need to know
delays, introduced by Speex algorithms. In particular, I'm interested
in delays of:
1) encoder in NB, WB and UWB modes,
2) decoder (I guess it should be 0?)
3) resampler. I think it varies depending on input rate to output rate
ratio. Any generic formula or at least values for specific ratios available?
4) AEC
5) preprocessor

If anyone knows delays added by at least some of these algorithms,
please, share this knowledge. I believe that this is worth doing and
I promise to compile a patch to Speex documentation with obtained
values and formulas. In VoIP we fight for every ms of latency, so lets
make people more aware of the actual values.

-- 
Regards,
Alexander Chemeris.

SIPez LLC.
SIP VoIP, IM and Presence Consulting
http://www.SIPez.com
tel: +1 (617) 273-4000

Alexander Chemeris a ?crit :
> I want to do a careful measure of mic-to-net and net-to-spkr delays
> in sip media engine and to make it really precise I need to know
> delays, introduced by Speex algorithms. In particular, I'm interested
> in delays of:
> 1) encoder in NB, WB and UWB modes,
NB: 25 ms, WB: ~28 ms, UWB: ~30 ms
> 2) decoder (I guess it should be 0?)
5 ms
> 3) resampler. I think it varies depending on input rate to output rate
> ratio. Any generic formula or at least values for specific ratios
available?
The higher the quality, the higher the delay
> 4) AEC
0 ms
> 5) preprocessor
1 frame extra delay (i.e. you get back the last frame you sent the last
time)

	Jean-Marc

Thank you for clarification. I attached a patch which (IMHO)
makes it more clear for everyone who just read Doxygen
docs. Resampler latency docs wording was just ambiguous
IMO, so I made it more robust to fast reading.

On Wed, Jul 1, 2009 at 5:16 PM, Jean-Marc
Valin<jean-marc.valin at usherbrooke.ca> wrote:
> Alexander Chemeris a ?crit :
>> On Tue, Jun 30, 2009 at 4:38 AM, Jean-Marc
>> Valin<jean-marc.valin at usherbrooke.ca> wrote:
>>> Alexander Chemeris a ?crit :
>>>> 2) decoder (I guess it should be 0?)
>>> 5 ms
>>
>> Hum, interesting. So, e.g. for NB mode delay is
>> 20 (packet size) + 5 (encoder lookahead) + 5 (decoder delay) = 30,
right?
>
> Correct
>
>>>> 3) resampler. I think it varies depending on input rate to
output rate
>>>> ratio. Any generic formula or at least values for specific
ratios available?
>>> The higher the quality, the higher the delay
>>
>> Seems we need to update our Speex sources. Our version has no
>> speex_resampler_get_[out|in]put_latency() functions. Should I
>> sum values of input and output latency to get overall latency?
>> Actually, why do you separate input and output latency? For
>> a user, resampler is a black-box, which logically has a single
>> latency value. Two latencies are confusing.
>
> You don't sum them, you just get the latency in different units.
> speex_resampler_get_input_latency() will return the latency in input
> samples, while the other will give you the same value, but expressed as
> a number of output samples.
>
> ? ? ? ?Jean-Marc
>


-- 
Regards,
Alexander Chemeris.

SIPez LLC.
SIP VoIP, IM and Presence Consulting
http://www.SIPez.com
tel: +1 (617) 273-4000
-------------- next part --------------
A non-text attachment was scrubbed...
Name: speex-latency-comments.diff
Type: text/x-diff
Size: 1458 bytes
Desc: not available
Url :
http://lists.xiph.org/pipermail/speex-dev/attachments/20090701/90366ff0/attachment.diff

First patch fixes two typos in manual.

Second patch adds few notes about latency, as promised.
Hopefully I haven't added any typos and grammar errors :)

On Wed, Jul 1, 2009 at 8:02 PM, Alexander
Chemeris<Alexander.Chemeris at sipez.com> wrote:
> Thank you for clarification. I attached a patch which (IMHO)
> makes it more clear for everyone who just read Doxygen
> docs. Resampler latency docs wording was just ambiguous
> IMO, so I made it more robust to fast reading.
>
> On Wed, Jul 1, 2009 at 5:16 PM, Jean-Marc
> Valin<jean-marc.valin at usherbrooke.ca> wrote:
>> Alexander Chemeris a ?crit :
>>> On Tue, Jun 30, 2009 at 4:38 AM, Jean-Marc
>>> Valin<jean-marc.valin at usherbrooke.ca> wrote:
>>>> Alexander Chemeris a ?crit :
>>>>> 2) decoder (I guess it should be 0?)
>>>> 5 ms
>>>
>>> Hum, interesting. So, e.g. for NB mode delay is
>>> 20 (packet size) + 5 (encoder lookahead) + 5 (decoder delay) = 30,
right?
>>
>> Correct
>>
>>>>> 3) resampler. I think it varies depending on input rate to
output rate
>>>>> ratio. Any generic formula or at least values for specific
ratios available?
>>>> The higher the quality, the higher the delay
>>>
>>> Seems we need to update our Speex sources. Our version has no
>>> speex_resampler_get_[out|in]put_latency() functions. Should I
>>> sum values of input and output latency to get overall latency?
>>> Actually, why do you separate input and output latency? For
>>> a user, resampler is a black-box, which logically has a single
>>> latency value. Two latencies are confusing.
>>
>> You don't sum them, you just get the latency in different units.
>> speex_resampler_get_input_latency() will return the latency in input
>> samples, while the other will give you the same value, but expressed as
>> a number of output samples.
>>
>> ? ? ? ?Jean-Marc
>>
>
>
>
> --
> Regards,
> Alexander Chemeris.
>
> SIPez LLC.
> SIP VoIP, IM and Presence Consulting
> http://www.SIPez.com
> tel: +1 (617) 273-4000
>


-- 
Regards,
Alexander Chemeris.

SIPez LLC.
SIP VoIP, IM and Presence Consulting
http://www.SIPez.com
tel: +1 (617) 273-4000
-------------- next part --------------
A non-text attachment was scrubbed...
Name: speex-manual-typos.diff
Type: text/x-diff
Size: 897 bytes
Desc: not available
Url :
http://lists.xiph.org/pipermail/speex-dev/attachments/20090701/1a2e4703/attachment-0002.diff
-------------- next part --------------
A non-text attachment was scrubbed...
Name: speex-manual-addons.diff
Type: text/x-diff
Size: 2838 bytes
Desc: not available
Url :
http://lists.xiph.org/pipermail/speex-dev/attachments/20090701/1a2e4703/attachment-0003.diff