[Ffmpeg-devel] FLAC encoder

[Corey Hickey]

Justin Ruggles wrote:
> Here are some current results.  The latest version is here:
> 
> http://home.earthlink.net/~jruggle/flake/flake-02.tar.bz2

You left your -march=k6-2 line uncommented in the Makefile. ;)

> If anyone wants to do tests with the clip I used for the test below:
> http://home.earthlink.net/~jruggle/flake/answering_bell.flac

I ran the test on my machine. I'm getting very slightly different
encoded sizes from yours. Here's the source file I used:

$ wget http://home.earthlink.net/~jruggle/flake/answering_bell.flac
$ flac -d answering_bell.flac -o test.wav
$ md5sum answering_bell.flac test.wav
79ab5c80142cce1478def3986ac6d5ac  answering_bell.flac
842d62cb88cea25d265f48be82eb2b5f  test.wav


...and here are the results (on amd64):


flac -f test.wav -o flac-
level   user time     bytes    ratio   kbps
-----  -----------  ---------  -----  -------
  0     0m0.216s     3257761   0.615   848.3
  1     0m0.228s     3249348   0.613   846.1
  2     0m0.335s     3233404   0.610   842.0
  3     0m0.331s     2998725   0.566   780.9
  4     0m0.293s     2957151   0.558   770.0
  5     0m0.384s     2942739   0.555   766.3
  6     0m0.517s     2940071   0.555   765.6
  7     0m2.372s     2933280   0.554   763.8
  8     0m3.154s     2920936   0.551   760.6

flake/flake test.wav flake-
level   user time     bytes    ratio   kbps
-----  -----------  ---------  -----  -------
  0     0m0.494s     3414019   0.644   889.0
  1     0m0.609s     3209480   0.606   835.8
  2     0m0.485s     3012291   0.568   784.4
  3     0m0.484s     2854938   0.539   743.4
  4     0m0.466s     2777988   0.524   723.4
  5     0m0.529s     2786305   0.526   725.6
  6     0m0.550s     2674994   0.505   696.6
  7     0m0.521s     2618914   0.494   682.0
  8     0m0.836s     2584388   0.488   673.0

**Note: don't pay too close attention to my user times: my computer is
busy doing other things and many of the small differences are within the
margin of error. The general trends toward higher times at higher levels
and flake being much faster at higher levels still hold true when I test
a longer file.

$ md5sum flac-test-* flake-test-*
26dcb63ac0a19a26391ae20868e9c57b  flac-test-0.flac
5ad4018b7c10d4b68b8c493c97044003  flac-test-1.flac
af3bd295afd9b176ab0a9d00128681c3  flac-test-2.flac
0e505cc0e6ddfe3b51ede6951ffd5d51  flac-test-3.flac
d067a1c99da33b7cbe8c5d9c55137c34  flac-test-4.flac
a7efc878330ddc43bb4d65f40abd5e92  flac-test-5.flac
2fb3b077f3d6255f37494aafc5ba8a72  flac-test-6.flac
174e06952f15fdf95404b261ee3380ef  flac-test-7.flac
488b786fa0b5758f3e44835bc3acaa46  flac-test-8.flac
d02d54a6a8f5c67030991cca9c8dad53  flake-test-0.flac
7a719c980fb2c336caf636eec1052683  flake-test-1.flac
b57ccd8a9b97cd630e0843ffc10cd830  flake-test-2.flac
8127f7bd2f62b2792767186cb022580a  flake-test-3.flac
0e5e94e7d8f9a2b96b6d8a3ad27debc0  flake-test-4.flac
c9242a4d0b9f348b6654b36534d92b31  flake-test-5.flac
bbced5bb5a2f1c1ce84f2245c1f79aa4  flake-test-6.flac
66041a63cc51e0d5b247ad3103b7afe1  flake-test-7.flac
3d572e704c313b8836a3561b2919de88  flake-test-8.flac


In any case, all the files sound fine. Every single file, when decoded
with "flac -d", ends up identical to the original wav. Good work!


For a laugh, or perhaps if anyone is masochistic, here's the bash mess I
used (attached so the lack of line breaks is preserved).

-Corey
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: messy
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20060601/0ef7476a/attachment.txt>