Flac - Sep 2004 - [Flac-users] flac 1.1.0 unable to encode 11025hz file

Early this past week, Miroslav Lichvar suggested for me:
> Ok, you need 0.04% improvement, that should not be a problem. Try
> flac --lax -e -p -l 32 -r 10 --no-padding
Thank you again, Miroslav.  I tried that, and it took almost two full 
days (surprisingly, Windows ME stayed up that long without crashing) to 
re-encode the entire set on my 266-MHz machine.  After all, in the help 
file Josh gives us fair warning that a couple of those options are slow. 
  But thirty-eight of the forty-six tracks came out larger than they had 
in my original attempt at the -8 preset (both groups were compared after 
stripping all metadata and padding except STREAMINFO and SEEKTABLE).

When I took the smaller version of each track, the total was still too 
big for a CDR without overburning -- by apparently less than 10 KB, but 
still too big.

Miroslav concluded,
> and if it is not enough, increase -r up to 16.
... so I tried -r 16 on the eight tracks that had benefited before.  All 
of the first four came out larger at -r 16 than at -r 10, and the 
detailed help display from flac --explain says that setting -r "above 4 
usually doesn't help much," so I stopped the process there.

In the end, I tried zipping the original .flac files done at -8, and 
although about sixteen (or fourteen?) of them came out larger, the total 
decreased enough to fit the music and .jpegs onto one CDR.  I tried 
removing the ones that had increased from the .zip archive, but neither 
Explorer nor WinZip could manage that without hanging.  Then I was going 
to write down which ones got larger from zipping and re-zip the others 
without them, but in the end I figured it wasn't worth the trouble to 
keep track of which ones belonged in the .zip archive and which didn't. 
  As long as the data fit onto one disc, that was good enough.

On Sat, Apr 12, 2003 at 05:49:06PM -0500, David W. Tamkin
wrote:
> Early this past week, Miroslav Lichvar suggested for me:
> 
> >Ok, you need 0.04% improvement, that should not be a problem. Try
> >flac --lax -e -p -l 32 -r 10 --no-padding
> 
> Thank you again, Miroslav.  I tried that, and it took almost two full 
> days (surprisingly, Windows ME stayed up that long without crashing) to 
> re-encode the entire set on my 266-MHz machine.  After all, in the help 
> file Josh gives us fair warning that a couple of those options are slow. 
Ah, sorry for wasting your time and electricity.
>  But thirty-eight of the forty-six tracks came out larger than they had 
> in my original attempt at the -8 preset (both groups were compared after 
> stripping all metadata and padding except STREAMINFO and SEEKTABLE).
Larger? Weird. Options of coding examined with suggested switches are
superset of options examined with -8, so i simply can't believe that
it is possible :). I have tried few files, but all were smaller than
with -8. Can you send me one small track which came out larger? 
> When I took the smaller version of each track, the total was still too 
> big for a CDR without overburning -- by apparently less than 10 KB, but 
> still too big.
> 
> Miroslav concluded,
> 
> >and if it is not enough, increase -r up to 16.
> 
> ... so I tried -r 16 on the eight tracks that had benefited before.  All 
> of the first four came out larger at -r 16 than at -r 10, and the 
> detailed help display from flac --explain says that setting -r "above
4
> usually doesn't help much," so I stopped the process there.
This should not be larger, really.

[...]

-- 
Miroslav Lichvar

--- "David W. Tamkin" <dattier@panix.com>
wrote:
> Early this past week, Miroslav Lichvar suggested for me:
> 
> > Ok, you need 0.04% improvement, that should not be a problem. Try
> > flac --lax -e -p -l 32 -r 10 --no-padding
> 
> Thank you again, Miroslav.  I tried that, and it took almost two full
> 
> days (surprisingly, Windows ME stayed up that long without crashing)
> to 
> re-encode the entire set on my 266-MHz machine.  After all, in the
> help 
> file Josh gives us fair warning that a couple of those options are
> slow. 
>   But thirty-eight of the forty-six tracks came out larger than they
> had 
> in my original attempt at the -8 preset (both groups were compared
> after 
> stripping all metadata and padding except STREAMINFO and SEEKTABLE).
> 
> When I took the smaller version of each track, the total was still
> too 
> big for a CDR without overburning -- by apparently less than 10 KB,
> but 
> still too big.
> 
> Miroslav concluded,
> 
> > and if it is not enough, increase -r up to 16.
> 
> ... so I tried -r 16 on the eight tracks that had benefited before. 
> All 
> of the first four came out larger at -r 16 than at -r 10, and the 
> detailed help display from flac --explain says that setting -r "above
> 4 
> usually doesn't help much," so I stopped the process there.
There is a bug somewhere related to the -r option, where the
file can end up bigger even when using a large -r value.  But
since the benefit is marginal I have not tracked it down.

Josh


__________________________________
Do you Yahoo!?
The New Yahoo! Search - Faster. Easier. Bingo.
http://search.yahoo.com

> Ah, sorry for wasting your time and electricity.
Seriously, thank you for the advice.  The waste was strictly my own.  I 
thought of zipping the JPGs before I posted the question, so why didn't 
I think of zipping the FLACs until later?

 > I have tried few files, but all were smaller than with -8.  Can you
 > send me one small track which came out larger?

Sorry, I no longer remember which they were.  The tracks that did better
under the options you suggested than at the -8 preset tended to be among
the smallest, and it was the larger ones that fared more poorly, so I'd
have to send you some of the larger ones, and my uplink is not that wide
(I'm lucky to send a meg a minute).

I had said,

 >> ... so I tried -r 16 on the eight tracks that had benefited before.
 >> All of the first four came out larger at -r 16 than at -r 10 ...

and Miroslav respnded,

 > This should not be larger, really.

Maybe I did something wrong?

The same situation arose yesterday, but the longest track was entirely
in off-center mono.  I converted it with GoldWave to straight mono
(which is less annoying besides), and then the FLACs and the JPGs all
fit easily, and there no need for zipping this time.

When I tried to encode a 11025hz wav file, I got the following error.  I 
have tried both 8-bit and 16-bit, mono and stereo.

-- Chuck

-----
D:\test>flac 11-16.wav

flac 1.1.0, Copyright (C) 2000,2001,2002,2003 Josh Coalson
flac comes with ABSOLUTELY NO WARRANTY.  This is free software, and you are
welcome to redistribute it under certain conditions.  Type `flac' for
details.

options: -P 4096 -b 4608 -m -l 8 -q 0 -r 3,3

11-16.wav: ERROR initializing encoder
            state = 14:FLAC__STREAM_ENCODER_NOT_STREAMABLE

On Thu, May 01, 2003 at 12:20:07PM -0500, Chuck H. Zhao
wrote:
> When I tried to encode a 11025hz wav file, I got the following error.  I 
> have tried both 8-bit and 16-bit, mono and stereo.
The sample rate is not in the Subset. Run flac with --lax option,
and it will be ok.

-- 
Miroslav Lichvar