[FFmpeg-soc] ALAC Testing Report

[Jai Menon]

Hi Mike,

On Monday 09 Jun 2008 11:57:12 am Mike Melanson wrote:
> Hi,
>
> At Jai's behest, I have started to test the ALAC encoder. I am using a
> test suite consisting of 13 "open source" music tracks that come from a
> CD called OpenMusic that I received at LinuxTag 2007. I actually like 2
> of the songs. Plus, they present varying types of music and sounds. I
> will post a link as soon as I can get all the raw CD rips uploaded to my
> server.
>

Thanks a lot for taking time off to do this! Really appreciate that...
 
> Anyway, here's the first report:
>
> 02-JeffreyAltergott-IcarusGrounded.wav
>        original WAV file size: 46,673,132 bytes
> Jai-2008-06-08 compresses to: 41,669,860 bytes
>          iTunes compresses to: 29,451,331 bytes
>
> So, clearly, the compression factor is a work in progress. That's
> understood. For the interoperability report:
>

Actually, the current svn doesn't include any predictor, not even a simple 
first order predictor. I have that working in my local repo and It will be 
reviewed by Justin very soon. With that, the compression is expected to be 
around 34-35 Mb for that file (just an estimate). We'll reach(surpass ?)  
iTunes compression after the generalised predictor is done.

> works in Windows/QT Player 7.4.5 (latest)
> works in Windows/iTunes 7.6.2.9 (latest)
> *doesn't work* on my iPod Touch
>

My guess is that has to do with the container more than the ALAC stream. The 
dbPoweramp author faced similar problems and it was related to muxing of the 
ALAC stream into an mp4 container. I'll surely investigate the issue further.

> Methodology:
>
>    ffmpeg -i <source.wav> -acodec alac dest.mov
>
> Problem: Specifying either dest.mp4 or dest.m4a does not work. It
> should. ALAC is supposed to be encapsulated in these formats and not in
> .mov, although that obviously works for iTunes and QT Player. I don't
> think the iPod firmware understands ALAC in MOV (although iTunes
> cheerfully transferred it).
>
I agree it should support mp4, but iirc the atom structure is non-standard. 
But once the encoder itself is done, I could see if we can get the mov muxer 
to work with this.

> Great start. The rest of you: Let the group know when your stuff is
> ready to get pounded on.

Thanks, and I hope we can continue this thread when the next major commit is 
in place.

Regards,

Jai Menon
<realityman at gmx.net>