From: Richard Ash <richard_at_audacityteam.org>
Date: Sat, 03 Sep 2011 12:47:39 +0100

On Thu, 2011-09-01 at 21:45 +0100, Jon Nicoll wrote:
> Hi there
> I have some m4a files which won't play properly on my Rockbox Sanza Fuze
> (V1). They seem to start off playing, and then a few minutes in, stop or jump
> to the next file. The files play fine on (for instance) VLC. I re-created them
> with MP4Box, which didn't help.

> (They originate via the 'get_iplayer utility BTW, which uses flvstreamer and
> FFmpeg).

How long are these files? I've had problems with long files on a
low-memory player (Samsung YH-925), which affected the real machine but
not the emulator. A bunch of fixes then made their way in through
flyspray #8923
 http://www.rockbox.org/tracker/task/8923

Since then I have lost the Samsung and got an iRiver iHP120, which
(running Rockbox 3.8.1) plays anything I have downloaded with
get_iplayer for it.

I'm not using the decoding built into get_iplayer however, because it
didn't work at the time I first set it up. So I have set get_iplayer to
emit raw files (.flv) then convert them with

 ffmpeg -vn -i "${flvfile}" -vn -acodec copy "${tmpfile}"

followed by (for files less than 2 hours)

 MP4Box -new -hint -add "${tmpfile}" "${outfile}"

If the files are more than two hours long I split them into 2-hour
pieces using MP4Box to avoid potential problems with large seek tables

  MP4Box -new -hint -add "${tmpfile}" -split ${max_duration} "$
{outfile}"

The result is outfile_001.mp4, outfile_002.mp4 and so on. FFmpeg and
MP4Box are current Gentoo amd64 stable versions, ffmpeg-0.7.3 and
gpac-0.4.5-r4

The full script is rather more complex because I check that the files
are the right size before deleting them, and then push the files out to
the player. I've attached the full script in case anyone is interested,
it seems to be fairly robust, and for one-off programs, you can download
them into the working directory with the right name suffixes and it will
convert and sync them.

As most of the proms recordings are well over the two hours mark (even
through the concerts are shorter than this, with the first and second
halves in separate programs), I wouldn't be surprised if large seek
tables are the problem. This would explain why the emulator works - with
lots of memory available in the host PC, I don't think malloc is
restricted in the same way it is on the actual player target.

Richard


-------------------------------------------------------------------
Unsubscribe: http://cool.haxx.se/cgi-bin/mailman/listinfo/rockbox
FAQ: http://www.rockbox.org/twiki/bin/view/Main/GeneralFAQ
Etiquette: http://www.rockbox.org/mail/etiquette.html

Received on 2011-09-03