[FFmpeg-devel] [PATCH] Common ACELP code & G.729 [6/7] - G.729 postfilter

[The Wanderer]

Diego Biurrun wrote:

> On Sat, May 17, 2008 at 11:55:32PM +0700, Vladimir Voroshilov wrote:
 >
>> 2008/5/17 The Wanderer <inverseparadox at comcast.net>:

>>> If I understand this sentence correctly, I would write it as something
>>> like
>>>
>>> ==
>>> At the second stage, the long-term postfilter searches for the
>>> best fractional pitch delay which has a resolution of 
>>> 1/(1+ANALYZED_FRAC_DELAYS) and is near the integer pitch delay
>>> which was  found in the first stage.
>>> ==
>> 
>> If i understood you right above sentence is not what i wanted to
>> say. In my sentence I just meant that search is being done in
>> [best_pitch_delay-1; best_pitch_delay+1] range with
>> 1/(1+ANALYZED_FRAC_DELAYS) precision. Where best_pitch_delay is the
>> integer pitch delay which was found in the first stage.
> 
> What about the following:
> 
>   In the second stage, the long-term postfilter searches for the best
>   fractional pitch delay with 1/(1+ANALYZED_FRAC_DELAYS) resolution
>   around the value found in the first stage.

This sounds to me like a fairly close rephrasing of what I wrote, which
Vladimir said was not correct.

Vladimir's own explanation sounds very similar to my understanding (and
thus to what my rephrasing was intended to mean), with one exception.
Vladimir, is it correct that 1/(1+ANALYZED_FRAC_DELAYS) is the precision
with/resolution to which the search is being done, rather than being
anything about the thing being searched for?

If that is correct, then (using Diego's phrasing as a base) the
grammatically correct way of saying it would be something like

==
In the second stage, the long-term postfilter performs a search with a 
precision of 1/(1+ANALYZED_FRAC_DELAYS) for the best fractional pitch 
delay around the value found in the first stage.
==

but it might be better to just omit the mention of the precision
entirely, i.e. drop the "with a precision of" etc. from the middle of
the above sentence. That has the downside of not mentioning the explicit
value being used, which may for all I know be important, but it is less
clunky to say in English.

Is there some reason why it would be necessary to inform the reader in
the comment of the precision being used?

-- 
       The Wanderer

Warning: Simply because I argue an issue does not mean I agree with any
side of it.

Secrecy is the beginning of tyranny.