From: Bluechip <csbluechip_at_gmail.com>
Date: Sat, 05 Nov 2005 09:30:21 +0000

I'm just looking at some Rockbox generated MP3's and have a couple of
questions I'd like to throw at any devs here.

1. Why does RB append a blank 4K ID3v2 header to the start of every file?
I can see the debate for prepending a v2 tag, although I personally suggest
it serves no useful purpose ...But if it is to be there, why not put some
data in it, not least of all the date and time?

2. Each ID3v2 tag is followed by 212 (0xD4) 0's ...which afaict serve no
purpose.
Is this a bug, or is there something I don't know?

3. There is next a corrupt Xing (VBRI) header containing the text "Rockbox
- rocks your box"
Again, is this helpful? Why not put the text in the tag? I presume VBRFix
will blat out the comment when you try to make the file useable.

4. The shift: it would seem from the four sample files I have here:
one of them shifts right by one bit 94 seconds in (frame 3655) ..that is
one extra bit has been suppended to frame 3654. (or debateably prepended to
frame 3655)
Two others have errors at the start and all have a LOT of extra 0's before
frame-1
a = 208bits of extra data (remains byte aligned)
b = 240bits of extra data (remains byte aligned)
The last one has a 115 bytes of garbage data, followed by 97 bytes of 0
..which looks suspiciously like the tail end of an MP3 frame.

I think the bug in the last three files is NOT a DAC problem, but a Rockbox
problem - furthermore I also suspect that the two files with a bunch of 0's
are actually the same problem as the third file, but it was not obvious
because the end of most frames is a bunch of 0's

It strikes me that this is most likely a bug in the time-split feature,
although my recorder almost never has this problem (once EVER - and I can't
say for sure what the problem was, I ignored it as a one-off) and I have
dne a LOT of recording in my time both with and without timesplit enabled -
so I am reliant on other supplying me corrupt files.

Bluechip
Received on 2005-11-05