Rockbox

Tasklist

FS#9776 - WMA Decoder - Problematic Files with example

Attached to Project: Rockbox
Opened by Siaoru (Siaoru) - Saturday, 10 January 2009, 12:06 GMT
Last edited by Andree Buschmann (Buschel) - Thursday, 03 February 2011, 11:06 GMT
Task Type Bugs
Category Codecs
Status Closed
Assigned To No-one
Operating System SW-codec
Severity Low
Priority Normal
Reported Version Daily build (which?)
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Dear all :)
I find some problem files for WMA decoder in the attached.

115.I.COULD.BE.THE.ONE.wma fails to play correctly in the beginning.
this file format as following:
Windows Media Audio 9.2
bitrate : 5 kbps, sample rate : 8 kHz, mono (A/V) 1-pass CBR

07.Devil.In.My.Mind.wma, I hear some noises.
This file format as following:
Windows Media Audio 9.2
bitrate : 32 kbps, sample rate :44.1 kHz, stereo 1-pass CBR

Could you fix these bugs for WMA codec?

Thanks you!
This task depends upon

Closed by  Andree Buschmann (Buschel)
Thursday, 03 February 2011, 11:06 GMT
Reason for closing:  Works For Me
Additional comments about closing:  Both attached files work fine using r29209.
Comment by MichaelGiacomelli (saratoga) - Saturday, 10 January 2009, 20:01 GMT
Neither of those files decodes quite right in ffmpeg, so fixing them is probably beyond me (though you may want to report them to ffmpeg to see if they have any ideas).

That said:

115.I.COULD.BE.THE.ONE.wma blows up while parsing the headers, but the same code works in ffmpeg, so its probably some sort of difficult to trouble shoot parser problem.

07.Devil.In.My.Mind.wma seems to have some sort of overflow from time to time. I will try to figure out whats wrong with this file.
Comment by Siaoru (Siaoru) - Sunday, 11 January 2009, 12:05 GMT
Dear MichaelGiacomelli,
About file 115.I.COULD.BE.THE.ONE.wma, I find that it fail because (s->block_pos + s->block_len) > s->frame_len when decoding wma_decode_block( ),
but it return success when goes to asf_parse_header( )
Is the unstandard format bitstream lead to this situation ?

Thank you very much!
Comment by MichaelGiacomelli (saratoga) - Sunday, 11 January 2009, 16:20 GMT
>About file 115.I.COULD.BE.THE.ONE.wma, I find that it fail because (s->block_pos + s->block_len) > s->frame_len when decoding wma_decode_block( )

I've seen this problem before. I have no idea what causes it, but I assume it is because we are not properly feeding the decoder with the right data. The condition errors out because its not allowed to have block that spans a frame boundary. Since ffmpeg uses the same code here, and does not error out, I assume it means there is a problem in the rockbox asf parser (which is not based on ffmpeg).

Comment by Siaoru (Siaoru) - Wednesday, 21 January 2009, 06:13 GMT
Dear MichaelGiacomelli,
I find out the bug of the file "115.I.COULD.BE.THE.ONE.wma", it has an informal replicated length in asf_read_packet.
Skip one byte if replicated_length equal to one!
Detail in the attached.

Regards,
Siaoru
   wma.c (19.1 KiB)
Comment by MichaelGiacomelli (saratoga) - Wednesday, 21 January 2009, 15:40 GMT
Great work!

If you give me your name I can credit you in the SVN commit.
Comment by Siaoru (Siaoru) - Thursday, 22 January 2009, 01:48 GMT
My name is "Siaoru" in the Rockbox.

Thank you!
Comment by MichaelGiacomelli (saratoga) - Thursday, 22 January 2009, 03:22 GMT
If you want credit you'd have to give me your real name, not the user name on this tracker.
Comment by Siaoru (Siaoru) - Sunday, 08 February 2009, 07:34 GMT
My real name is "Siaoru Lee", but I just register it on Rockbox only. How to register it on SVN ?
Thank you !
Comment by Siaoru (Siaoru) - Sunday, 08 February 2009, 08:13 GMT
My real name is "Siaoru Lee", but I just register it on Rockbox only. How to register it on SVN ?
Thank you !
Comment by MichaelGiacomelli (saratoga) - Sunday, 08 February 2009, 22:17 GMT
Thank you. I've committed a simple note with the fix saying that it was your work and not mine:

http://svn.rockbox.org/viewvc.cgi?view=rev;revision=19946

Comment by Dave Hooper (stripwax) - Sunday, 18 October 2009, 02:27 GMT
saratoga - do you think the (potential) problem with the first file (that looks like an asf parsing issue) is the same as  FS#10027  ?
Comment by MichaelGiacomelli (saratoga) - Sunday, 18 October 2009, 03:22 GMT
I'm not sure. I don't actually know anything about the ASF parser (linuxstb wrote it). I suppose if you're interested in finding out you could use MS's "asf view" and compare the rockbox parser output to MS's tool. Sitting down with it and learning how ASF works has been on my todo list for ages but not very near the top.

Loading...