> From: "Mathew Hendry" <[EMAIL PROTECTED]>
> Date: Sun, 28 Nov 1999 05:17:06 -
>
> Now with Takehiro's table changes; in iteration_init:
>
wow, that was fast!
I just committed the new routines to sourceforge.net. They
make a *big* difference, at least under gcc.
I also took Takehiro'
Now with Takehiro's table changes; in iteration_init:
=
#if (defined(__GNUC__) && defined(__i386__)) || defined (_MSC_VER)
for (i = 0; i < PRECALC_SIZE - 1; i++)
adj43[i] = i + 0.5 - pow(0.5 * (pow43[i] + pow43[i + 1]), 0.75);
adj43[i] = 0.0;
#else
for (i = 0; i < PRECALC
Hi all,
This new version gives approx 70% overall speed increase under MSVC; gcc
version is untested but should be okay, or at least trivial to fix. ;)
We could get a small additional speed increase by rolling some of the "-
0.5"s into adj43[].
#define XRPOW_ASM
#ifdef XRPOW_ASM
# if defined(_
Hi, all.
>>One problem: The code is now about 50% slower, mostly because
>>the old quantize_xrpow had some hand rolled asm for gcc and MSVC.
No problem :)
This is ASM implementation.
# I have no MSVC and am not good at MSVC, so there's no MSVC code.
# But it is easy to implement on MSVC, I hop
> On Sat, 27 Nov 1999, Mark Taylor wrote:
> > You should be able to get the latest version from anon cvs,
> > which has Segher's quantization fix. (and Takehiro's implementaton).
>
> How else does this differ from 3.57? It seems to be consistantly encoding
> with a high birate in VBR (same sett
On Sat, 27 Nov 1999, Mark Taylor wrote:
> You should be able to get the latest version from anon cvs,
> which has Segher's quantization fix. (and Takehiro's implementaton).
How else does this differ from 3.57? It seems to be consistantly encoding
with a high birate in VBR (same settings; -V4).
> From: Chuck Zenkus <[EMAIL PROTECTED]>
> Date: Sat, 27 Nov 1999 15:13:51 -0600
>
> Let's try that again!
>
> <> <>
That's great! I just added it to the web page, in the
'software that supports lame' section.
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
On Sat, 27 Nov 1999, Mark Taylor wrote:
>
> LAME source code is now on sourceforge.net :-) We still have 3 files
> with ISO code, so this may not be 100% legit, but it is source code
> only and lets try to keep it low profile!
You could impliment lame as a seperate branch to the ISO code, and
On Sat, 27 Nov 1999, Mark Taylor wrote:
>
> LAME source code is now on sourceforge.net :-) We still have 3 files
Works quite nicely. I just cvsupdated.
> You should be able to get the latest version from anon cvs,
> which has Segher's quantization fix. (and Takehiro's implementaton).
Work
LAME source code is now on sourceforge.net :-) We still have 3 files
with ISO code, so this may not be 100% legit, but it is source code
only and lets try to keep it low profile!
You should be able to get the latest version from anon cvs,
which has Segher's quantization fix. (and Takehiro's i
> From: "Renato C. Cerchiari" <[EMAIL PROTECTED]>
> Date: Fri, 26 Nov 1999 14:27:00 -0200
>
> What's happenning with VBR in Lame?
> Encoding a song with version 3.54 give me an average bitrate of 150 kbps.
> The same song with version 3.57 give me 124 kbps.
> Parameters are the same in both cases
Let's try that again!
<> <>
LOGO_2.jpg
LOGO_1.jpg
> From: "Gabriel Bouvigne" <[EMAIL PROTECTED]>
> Date: Sat, 27 Nov 1999 14:03:02 +0100
>
> > The midpoint (in terms of the error) is: .5 * (1 + 2.52) = 1.76.
> > So if x<1.76, we should take i=1 and if x>1.76 we take i=2.
>
> I think that for this kind of test, something like:
> i=(int)(x^.75+.
What's up with this?
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
- Original Message -
From: "Gabriel Bouvigne" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Saturday, November 27, 1999 1:03 PM
Subject: Re: [MP3 ENCODER] quantization formula
> > The midpoint (in terms of the error) is: .5 * (1 + 2.52) = 1.76.
> > So if x<1.76, we should take i=1
Hi everybody,
upgraded my computer, previously had a Pentium 90, 32 MB Ram, 3 HDs
(all SCSI, each 2.1 GB). Now have an AMD-K6-III-400, 192 MB Ram, 2 HDs
(U2W-SCSI, 18 GB and 9 GB). My previous NCR-53810 has been replaced by
a Tekram DC390-U2W. My Linux-Box got copied exactly the way it was on
the
> This 'downsampling' will have no beneficial effects on aliasing above simply
> dropping every other sample. Linear averaging is not the same as a lowpass
> filter.
dang. of course you are right.. i should have sepnt 5 seconds of thought on
the problem before starting to code right away... that
> And adj43[] is a lookup table, building up like this.
>
> for (i = 0; i < PRECALC_SIZE - 1; i++)
> adj43[i] = (i + 1) - pow(0.5 * (pow43[i] + pow43[i + 1]), 0.75);
>
> adj43[i] = 0.5;
> ---
> Takehiro TOMINAGA // may the source be with you!
> --
> MP3 ENCODER mailing list ( http
> "S" == Segher Boessenkool <[EMAIL PROTECTED]> writes:
S> Calculate i = floor(pow(x, 0.75)) as usual. i is integer.
S> Now:
S> if (x > pow(0.5 * (pow(i, 8.0/3.0) + pow(i+1, 8.0/3.0)), 0.375))
S> i++;
S> and i is the quantized value.
S> Rationale: minimize the
I sent a email to Mark, but I caught returned mail.
It says:
The original message was received at Sat, 27 Nov 1999 22:00:45 +0900 (JST)
from mail.ca2.so-net.ne.jp [210.139.254.22]
- The following addresses had permanent fatal errors -
<[EMAIL PROTECTED]> (reason: 550 Access deni
> The midpoint (in terms of the error) is: .5 * (1 + 2.52) = 1.76.
> So if x<1.76, we should take i=1 and if x>1.76 we take i=2.
I think that for this kind of test, something like:
i=(int)(x^.75+.5)
would be faster than a comparison, and produces the same output.
Btw do you have any explanatio
> Just to explain a little more what Segher is writing about:
> MP3 uses a nonlinear quantization of the MDCT coefficients.
> Let x = an MDCT coefficient after all scalefactors have been applied.
> MP3 needs to truncate x^(3/4) to an integer.
>
> As an example,
> x=1.80
> x^3/4 = 1.55
>
> two po
>
> Just to explain a little more what Segher is writing about:
> MP3 uses a nonlinear quantization of the MDCT coefficients.
> Let x = an MDCT coefficient after all scalefactors have been applied.
> MP3 needs to truncate x^(3/4) to an integer.
>
> As an example,
> x=1.80
> x^3/4 = 1.55
>
> tw
23 matches
Mail list logo
