-----Original Message-----
From: J. Droppo <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
Date: Wednesday, September 02, 1998 5:34 PM
Subject: Re: [MP3 ENCODER] trig functions
>The atan2() in fft.c computes the phase of each sample of the fft. The only
place
>this is used is in L3_psycho_analize(), where the cos() and sin() of this
angle are
>used, as well as the cos() and sin() of functions of this angle. Using trig
>identities
>
>sin(A-B) = sin(A) cos(B) - cos(A) sin(B)
>cos(A-B) = cos(A) cos(B) + sin(A) sin(B)
>sin(2A) = 2 cos(A) sin(A)
>cos(2A) = cos(A)**2 - sin(A)**2
>
>I re-formulated caluculations of the form cos(2*atan2(A,B)-atan2(C,D)) as
functions
>of A, B, C, and D, which are just the real and imaginary parts of one bin
of the
>DFT.
>
>Anyway, it seems rather complicated to explain now, but it does give a 4%
speed
>improvement over version "2b-m11" of the code I downloaded. Maybe someone
can take a
>look and speed up the calculation in L3_psycho_analize() even more? Search
for the
>constants m1, m2, and m3 to find the two major revisions.
>
>Gregory Hosler wrote:
>
>> I'd certainly be interested in those fixes, but of course they really
should
>> go back to the maintainer.
>>
>> On another note. Frequently I notice a "slight pop" at the very begining
of a
>> (most) track(s). I once saw someone suspect that this might be an
uninitialized
>> buffer, or variable. has anyone looked into this further, or run purify
on 8hz ?
>>
>> regards,
>>
>> -Greg
>>
>> On 01-Sep-98 "J. Droppo" wrote:
>> >Hi, I'm new to this list.
>> >
>> >I've noticed that the sources still have <yuck> unnecessary trig
>> >functions </yuck> in fft.c and l3psy.c. They can be removed from the 8hz
>> >02b sources, and gain about 10% speed without trying hard. My question
>> >is, assuming that speed is good and it hasn't been tried, who should I
>> >send the fix to?
>> >
>>
>> ----------------------------------
>> E-Mail: Gregory Hosler <[EMAIL PROTECTED]>
>> Date: 02-Sep-98
>> Time: 09:03:528
>>
>> "And where were you at 00:00:00 GMT, January 1, 1970?"
>>
>> --
>> As far as the laws of mathematics refer to reality, they are not
>> certain, and as far as they are certain, they do not refer to reality.
>> -- Albert Einstein
>> ----------------------------------
>>
>
fft.h
fft.c
l3psy.c
