First of all, sorry for my English, I'm French...

I had an original CD.
I extracted it in separated files per track (I get the timecode of each chapter).
Then I reassemble them into single flac file.
I muxed it into a chaptered MKA (with the eventual pb I explained here in French).

PROBLEM 1
The more important problem is that Foobar2000 crashes at the end of the last track (defined by the last chapter) when output method is "DirectSound v1/2", and Windows crashes when output method is "Kernel"...

It chrashed at the end of the last track:
- when playing,
- when replaygaining,
- when transcoding...

I think I know where is the problem:
- my original flac file has 138278196 samples (read in Foobar2000 file properties).
- my mka file has 138281415 samples (3219 more than the flac).
- when I re-extract flac from mka, it has 138278196 samples...
So I think when foobar is reading the 3219 virtual samples, it crashes...

PROBLEM 2
There's another problem:
When I read one track properties (defined by chapters) in the mka, for example, I have 976080 samples, and when I convert it into flac separated file (with flac transport layer), it has 976079 samples, so 1 sample is missing...
Not very much, but after that the album is not lossless!

I think the problem come from round time chapters (read here in French).
- 1/44100 has infinite of numbers after point, so the precision of nanosecond is an approximation...



So to resolve the problem, I think Foobar has to consider the approximation of nanosecond, considering the round is by default (6.1 --> 6 ; 6.9 --> 6, not 6.9 --> 7)



That would be great if these bugs can be fixed in futures versions

Thanks for your work

Maybe you'd be interested in reading this.

Try using 0.8.6 for muxing. Should make mka files that play in foobar.

Arf, 0.86 doesn't support tagging @ nanosecond (

The same file MKA doesn't crashes MediaPlayerClassic... So I think the problem come from a Foobar2000 bug

Is it possible for developpers to fix these bugs?

I'm waiting in suspense for the next beta

Foobar seems to crash at the end only if the muxed stream in mka is flac, it seems to not crash with ogg in mka ...

Yep, I experienced this as well. It only crashes on files containing FLAC. Well, I think FLAC is the only MKA contained audio format that is not handled by the foo_input_std component. So possibly the bug is not necessarily in foo_matroska but in the FLAC component??

I hope Peter will find the time to look into this, as mkvtoolnix is developing extremely fast and I don't like to stick with an extra version in the future.

tiki4

I tried to mux a flac encoded with an ogg transport layer, same problem...
And I saw another bug: I can't tag a flac with ogg as transport layer with foobar without crash...

And will the 2nd problem can be corrected? It will be great

The first pb is half-resolved (foobar doesn't crash anymore) with foo_matroska 0.6.

...

When foobar crashes, do you get an error log? If so, please post it.

[edit]
P.s. you might want to check out this thread as well: http://www.hydrogenaudio.org/forums/index....topic=24096&hl=
[/edit]

It doesn't crash anymore, with foo_matroska 0.6, but there are bugs in number of samples

Anyone tryied TTA (+CUE) in MKA? I have some problems with it. It even doesn't want to play in foobar. Tested on latest "mkvtoolnix-0.9.3-build20040723-1".
"ERROR (foo_matroska) : Matroska: unable to find a "A_TTA1" packet decoder object."

Regards
Lobuz

@Lobuz:

There is no packet decoder for TTA yet. This is a required feature to decode any format from Matroska in foobar2000. For now only MP2/3, AAC, Vorbis and FLAC are working (maybe also uncompressed PCM, don't know).

Regards,

tiki4

P.S.: Your TTA/MKA files should play in any DirectShow based player with the MatroskaSplitter and TTA DirectShow filters posted here recently.

Same problem with REPLAYGAIN: applying replaygain on mka doesn't make the same results (-8.93dB instead of -8.94dB for example)...

It's because of ms accuracy of seeking...

I'm waiting Mosu to make new mmg