David Byron wrote:
> I just did something a bit silly but I've got my fingers crossed that
> someone can help. I just did some recording using an ancient version of
> rockbox. The info screen on my Archos Jukebox Recorder 20 says version 2.1.
> I should have upgraded first. Or maybe folks here will tell me I should
> have bought a newer mp3 recorder but that's a bit of a different question.
As the Archos jukeboxes and recorders were the first devices supported
by rockbox, version 2.1 was pretty mature in all core functionalities
> In any case, what's done is done and I have an ~65 minute mp3 file that
> plays fine on the device but when I transfer it to my windows machine, I
> can't find a media player to play it all the way through.
You're probably hit by a bug in the hardware, the MAS MP3
encoding/decoding chip of the Archos boxes. Unfortunately, I couldn't
find any reference on the rockbox site about it, so everything I write
is just off my head about discussions on this list about five years ago.
Somebody found out that the MAS produces faulty recordings under certain
circumstances. AFAIR, exceeding the input level limit was one way to
increase the probability of the bug. When the bug happens, the MP3
stream of the recorded data contains a few extra BITS in the wrong
place. This makes all the following data look like garbage, as
everything is shifted a little bit...
> I'm looking for help decoding this file. I grabbed the rockbox source code
> and am looking into building the relevant bits of code for a desktop target.
> No idea if that's feasible or if the Archos hardware is doing the magic
> here. Please let me know if I should move this over to the developers list.
The Archos recorders are devices with hardware decoders, so you won't
find the code necessary to decode your file in the rockbox sources. The
most simple solution in your case would be to play the file on the
recorder and rerecord the audio on your computer. Not nice and not
perfect quality-wise, but easy to do.
It is possible to fix the broken recording however. You need to analyze
the MP3 stream, find the error in the bitstream and remove the
unnecessary bits. Someone (bluechip or Jens Arnold, I don't remember)
started to write a tool to do this automatically. I don't know what
happened to this tool, however...
This is such a strange and uncommon bug, that the classical mp3 fixing
tools don't work here, unfortunately.
Received on 2009-11-23