mkvtoolnix (Matroska toolkit): new release [Archive] - Page 6

jk888

5th January 2004, 16:27

Originally posted by Mosu
A lot of people have this problem, but so far I haven't been able to reproduce it. Therefore I need sample files. If you all could please upload VobSubs for which mkvmerge crashes then I could fix this sooner. My FTP server is always open: mosu.no-ip.com, user "upload", password "only". And please also upload a text file with some description and your name/nick name so that I can know who uploaded which file.

Thanks.

Hi Mosu I uploaded my subtitle files to you FTP.

Here is the command line from mkvmerge v0.80 build20040103:

"mkvmerge" -o "F:\DVD3\Memories of Murder.mkv" -d 0 -A -S --track-order 0 "F:\DVD3\test.avi" --aac-is-sbr 1 -a 1 -D -S --track-order 1 "F:\DVD3\test.aac" --language 0:kor --language 1:eng -s 0,1 -D -A --track-order 0,1 "F:\DVD3\VTS_01_0.idx" --title "Memories of Murder" --split "700M" --dont-link --chapters "F:\DVD3\chapters.txt"

The crash can happen anytime anywhere during the entire process, from the scanning process to the splitting, just anytime, all random. If I change the track order, or exlude something (eg chapters file) then the crash will happen in a different area, or if I'm lucky then everything works.

There's nothing wrong with the AVI, AAC or chapter files, I can mux
everything fine without the subtitles. Once I try to add the subtitles then it crashes. Most of the time when I mux the AVI, AAC files together then later mux subtitles into that mkv file then it works.

--------------------------------------------------
1st subtitle file: Korean, English

The 1st set of vobsubs I sent you also contain this error message:

Warning: vobsub_packetizer: Encountered broken SPU packet (next_off < start_off) at timecode 01:10:22.200. This packet might be displayed incorrectly or not at all.

I'm not sure if this is related to that particular subtitle, this is the first time I encountered this error. Maybe this subtitle is bugged. I have included the log file as well.

Also this subtitle contained colors in the middle of the movie.

--------------------------------------------------
2nd subtitle file: Chinese, English

The second set of subtitles never had any error messages on it. But ofcourse it still crashes when I mux it into my AVI + AAC movies. In fact every single one of my subtitles crash...

Mosu

6th January 2004, 01:21

Heya,

the second release in 2004 right after 0.8.0 is 0.8.1. It fixes three important things, and though I'm quite annoyed at having introduced the first with 0.8.0 in the first place I'm as pleased to have finally fixed the other two.

The first bug, introduced with 0.8.0, made muxing RealVideo impossible (resulting in segfaults / invalid memory access etc.). This bugfix changed two lines of code.

The second bug fixes the parsing of QuickTime/MP4 files for some special atom sizes. This mostly affects MP4 files created by NeroDigital and consists of eight lines of source code being moved down one line.

The third bug fixes the VobSub handling on Windows getting rid of these illegal memory access / "mkvmerge exited with an error -13905490871395". The bugfix is exactly one line long.

You see, small changes with big results.

The URLs:
http://www.bunkus.org/videotools/mkvtoolnix/
The source:
http://www.bunkus.org/videotools/mkvtoolnix/sources/mkvtoolnix-0.8.1.tar.bz2
And the Windows binaries:
http://www.bunkus.org/videotools/mkvtoolnix/win32/mkvtoolnix-0.8.1.rar

The ChangeLog:
-------------------------
2004-01-06 Moritz Bunkus <moritz@bunkus.org>
* Released v0.8.1.

2004-01-05 Moritz Bunkus <moritz@bunkus.org>
* mkvmerge: bug fix: The I/O classes were not initialized correctly on Windows resulting in spontaneous strange error messages, especially when muxing VobSubs.

2004-01-03 Moritz Bunkus <moritz@bunkus.org>
* mkvmerge: bug fix: For some special atom sizes in Quicktime and MP4 files the size was not read correctly. This affected e.g. files created by Nero Digital.

2004-01-02 Moritz Bunkus <moritz@bunkus.org>
* mkvmerge: bug fix: Segfault when muxing some video formats due to unchecked data (includes RealVideo).
----------------

Have a pleasant day :)

Rash

6th January 2004, 04:14

Originally posted by Mosu
You see, small changes with big results.
And you are the man! Thanks a lot Mosu. ;)

jk888

6th January 2004, 14:10

Hurray! Subtitles work now, flawlessly as well! I can't believe how quick you fixed it Mosu, many thx! I should have submited this bug 2 months ago, and all the while I was hoping the next versions would eventually fix it. From now on I'll submit bugs when I see one.

Mosu

6th January 2004, 14:34

Originally posted by jk888
Hurray! Subtitles work now, flawlessly as well! I can't believe how quick you fixed it Mosu, many thx! I should have submited this bug 2 months ago, and all the while I was hoping the next versions would eventually fix it. From now on I'll submit bugs when I see one.

That's definitely better ;) This bug is a perfect example why because it was located in the Windows specific code, and I only develop and test on Linux. So the Windows code doesn't get all that much attention.

ssjkakaroto

6th January 2004, 15:42

hey mosu all your work on mkvtoolnix has been superb :D
thanks a lot!

KpeX

6th January 2004, 16:22

Mosu, I'd like to thank you once again for developing on and supporting an OS that you don't even use personally with MKVtoolnix. I'm sure it can be a pain at times, but it is greatly appreciated for all of us Windows users, especially considering the features that MKVtoolnix has that no other tools support ATM. There are a lot of open-source projects that have nowhere near the Windows development and support that MKVtoolnix does, so thanks. And keep up those one line bugfixes ;).

JasonFly

7th January 2004, 18:13

Thanks Mosu for all your work.

About the vobsub bugfix. This one was very strange one because for me sometimes Vobsub muxing worked perfectly and sometimes I get that strange error message.
It has also happen that muxing worked once and doesn't work five minutes later with the same files. It was the second part of a .sub which has been cut. I just recut the original Sub and this time, muxing worked.Weird, isn't it?

I haven't tried 0.8.1 but I have great hope.

Thanks again.

Mosu

7th January 2004, 18:30

Originally posted by JasonFly
Thanks Mosu for all your work.

About the vobsub bugfix. This one was very strange one because for me sometimes Vobsub muxing worked perfectly and sometimes I get that strange error message.

Here's the technical explanation. In my I/O framework I have a general input/output class with two constructors, one with a filename and one without any arguments. Now the VobSub packetizer uses a derived class for input/output to a memory region which of course does not need a filename. Unfortunately the code for this constructor looked like this:

mm_io_c::mm_io_c() {
file = NULL;
safefree(file_name);
}

Now don't ask me why I wrote safefree() there - it's obviously total bullshit. It should read 'file_name = NULL;' instead. Now, depending on what the memory at file_name looked like, safefree would be lucky (e.g. if file_name pointed to NULL then safefree would just do nothing), but then again that memory might contain anything as it wasn't initialized resulting in a free() call to some weird address. That's why it happened only sometimes.

jk888

8th January 2004, 18:16

Originally posted by JasonFly

I haven't tried 0.8.1 but I have great hope.

Try it. Cause if you read the top it's the one that fixed it.

jk888

9th January 2004, 16:13

Hi Mosu,

Some of the sentences are missing for subtitles that are muxed into Matroska files using v0.8.1. I noticed this only after doing some extensive testing of the new mkvtoolnix.

For the 1st set of subtitles I sent you the korean/english subtitle set, the english sentences on these timestamps are missing:

09:26
12:39
13:55
15:23
21:01
.
.
.
etc
I have no idea if any Korean sentences are missing, because I can't read Korean.

Seems to happen randomly and occurs very often. The subtitle set displays correctly when it's not muxed in with the video, so I don't think there's anything wrong with the sub file.

Mosu

9th January 2004, 16:24

Originally posted by jk888
Hi Mosu,

Some of the sentences are missing for subtitles that are muxed into Matroska files using v0.8.1. I noticed this only after doing some extensive testing of the new mkvtoolnix.

I'm sure it's a playback problem and not a muxing problem. Run mkvinfo, e.g. 'mkvinfo -v your_matroska_file.mkv > output.txt', look through output.txt and check the timestamps you noticed missing subs for. If there ARE packets at those timestamps then it is a playback problem.

jk888

9th January 2004, 17:54

Mosu

9th January 2004, 17:57

Originally posted by jk888
Here's the results of my findings:
...

Mosu, do you want me to upload the output.txt file?

Yes please, and while you're at it, please also upload the subtitles in question to my FTP server (address: mosu.no-ip.com, user: "upload", password: "only"). Please compress the output.txt with an archiver of your choice as they tend to get very big ;)

Thanks.

jk888

9th January 2004, 18:01

Wow Mosu, talk about instant customer service hehehe, that was very quick reply.

I have uploaded the output.txt file into /vobsub-crash/1st sub/

Thx alot for looking into it for me :)

Mosu

9th January 2004, 18:16

Originally posted by jk888
Wow Mosu, talk about instant customer service hehehe, that was very quick reply.

I have uploaded the output.txt file into /vobsub-crash/1st sub/

Thx alot for looking into it for me :)

Ok, I just realized that you're still using a 0.8.0 build. Please download and test 0.8.1 (see my sig for a direct link) as it contains an important VobSub bugfix.

I'll look into the outputs now.

Mosu

9th January 2004, 18:21

Originally posted by jk888
From what I can tell by looking at this stuff, and I'm no programmer so correct me if I'm wrong, seems like the missing subtitles got the wrong track number. Track 1 (video) was used instead of Track 2 (English subtitle).

From what I can see there are subtitle packets for the timecodes you mentioned. Example:

| + Block group
| + Block (track number 1, 1 frame(s), timecode 566.332s)
| + Frame with size 1131
| + Reference block: -33.000ms
| + Block group
| + Block (track number 2, 1 frame(s), timecode 566.332s)
| + Frame with size 2562
| + Block duration: 5302.000ms
| + Block group
| + Block (track number 3, 1 frame(s), timecode 566.332s)
| + Frame with size 1398
| + Block duration: 5205.000ms

and

| + Block group
| + Block (track number 1, 1 frame(s), timecode 758.591s)
| + Frame with size 1929
| + Reference block: -33.000ms
| + Block group
| + Block (track number 2, 1 frame(s), timecode 758.591s)
| + Frame with size 692
| + Block duration: 2127.000ms
| + Block group
| + Block (track number 3, 1 frame(s), timecode 758.591s)
| + Frame with size 1265
| + Block duration: 2068.000ms

Yes, there's a video frame at this timecode, but there are frames for the sub tracks as well. Similar for the other timecodes you mentioned (although I haven't checked them all).

jk888

9th January 2004, 18:59

Very sorry Mosu, I should have been more careful reading those sections. Yes I noticed it now as well, the subtitle tracks shows up after the video track.

In that case I am not sure what went wrong. I am already using the latest VSfilter 2.32 from the Matroska Driver Pack 1.01. And it does work correctly when I play the movie without muxing the subititles in (eg. playing the movie with the sub files in the same directory with the same file name and VSfilter would auto load the subtitles).

Problems seem to appear only after it gets muxed into the mkv file. If I play a mkv file that doesn't have a subtitle track, then VSFilter would auto load the subtitle and everything displays correctly.

Also I am using 0.8.1, I double checked it and the GUI about says 0.8.1, but I redownloaded it and reinstalled it anyways, same problem appears.

jk888

9th January 2004, 19:31

I mentioned before that I get this error message when I muxed the subtitles for this particular movie.

Warning: vobsub_packetizer: Encountered broken SPU packet (next_off < start_off) at timecode 01:10:22.200. This packet might be displayed incorrectly or not at all.

Now does this have anything to do with the missing sentences? mkvtoolnix allows me to finish muxing the file even with this error.

Mosu

9th January 2004, 20:38

Originally posted by jk888
I mentioned before that I get this error message when I muxed the subtitles for this particular movie.

Warning: vobsub_packetizer: Encountered broken SPU packet (next_off < start_off) at timecode 01:10:22.200. This packet might be displayed incorrectly or not at all.

Now does this have anything to do with the missing sentences? mkvtoolnix allows me to finish muxing the file even with this error.

I just checked with mplayer on Linux, and mplayer prints a warning about the SPU packets not being correct. mkvmerge does not (or should not, maybe I really have produced a bug) alter the SPU packets, so I don't know what's up with them. I'll investigate, though.

Mosu

9th January 2004, 23:23

jk888: I think I've fixed the bug in mkvmerge. The problem was that it was copying SPU packets for the wrong stream on occasion. Please download and try this release: http://www.bunkus.org/videotools/mkvtoolnix/win32/pre/mkvtoolnix-0.8.1-build20040109-1.rar

Palikrovol

10th January 2004, 00:34

OK.

Now it works for me. At last the unseen subtitles are visible :).

I used the last build. I haven't try if with the first 0.8.1 build it also works.

Mosu, you are good Doc :D

Thank you vary much.

Palikrovol

10th January 2004, 01:01

I have just tried the first 0.8.1 build and some subtitles are not displayed (those that are visible muxing with version 0.8.1 build20040109-1) so the unseen subtitles bug seems to be fixed in the last build.

jk888: Please confirm it also works for you :)

Happy Day :)

Thanks again Mosu

Human_USB

10th January 2004, 03:54

I think I have found a bug but I need to test it more. I will tell you what I find.

-Jason

jk888

10th January 2004, 11:30

Hi Mosu, I tried the latest build, some subtitles that were missing are now visible, but some are still missing.

09:26 - visible
12:39 - visible
13:55 - missing *
15:23 - visible
21:01 - missing *

Also I don't get the broken SPU packet warning anymore when I mux. Maybe there are 2 diff bugs causing subtitles to disappear? I'm not too sure.

Mosu

10th January 2004, 12:15

Originally posted by jk888
Hi Mosu, I tried the latest build, some subtitles that were missing are now visible, but some are still missing.

09:26 - visible
12:39 - visible
13:55 - missing *
15:23 - visible
21:01 - missing *

Also I don't get the broken SPU packet warning anymore when I mux. Maybe there are 2 diff bugs causing subtitles to disappear? I'm not too sure.

Ok, thanks for the report, I'll look into those timecodes.

Mosu

10th January 2004, 14:00

Originally posted by jk888
Hi Mosu, I tried the latest build, some subtitles that were missing are now visible, but some are still missing.

09:26 - visible
12:39 - visible
13:55 - missing *
15:23 - visible
21:01 - missing *

Also I don't get the broken SPU packet warning anymore when I mux. Maybe there are 2 diff bugs causing subtitles to disappear? I'm not too sure.

Next try. Now for me all the items at the timecodes you mentioned show up. Please download and verify yourself: http://www.bunkus.org/videotools/mkvtoolnix/win32/pre/mkvtoolnix-0.8.1-build20040110-1.rar

Thanks.

jk888

10th January 2004, 19:38

I watched the movie from beginning to end, all the subtitles showed up. Finally this bug is squashed for good! :p

Thx Mosu for fixing this particularly annoying bug, you have great patience. I think the subtitles feature in mkvtoolnix is now fully tested and complete. *I breath a sign of relief*

Just wondering, can you provide details of the bug you fixed?

Mosu

10th January 2004, 19:50

Originally posted by jk888
I watched the movie from beginning to end, all the subtitles showed up. Finally this bug is squashed for good! :p

Thx Mosu for fixing this particularly annoying bug, you have great patience. I think the subtitles feature in mkvtoolnix is now fully tested and complete. *I breath a sign of relief*

Geee, then imagine how happy I am :)

Just wondering, can you provide details of the bug you fixed?

Sure. There were two bugs, actually. A little explanation:

As you probably know the .idx files contain index lines which tell where to find the MPEG packets for the subs for track n at the given timecode. Example:
timestamp: 00:03:58:520, filepos: 000019000

Unfortunately this does not tell us the length of this block. Now there are such lines for all tracks in a single .idx file, and the MPEG packets for these tracks may even be interleaved. A simplified example (simplified snytax, that is):

track 0
timestamp: 0, filepos: 0
timestamp: 1, filepos: 1000

track 1
timestamp: 0, filepos: 500
timestamp: 1, filepos: 1500

So I thought to myself: "Let's sort all these index lines by their filepos. Then use the difference between two such entries as the length for the first one." This would mean that the first block for track 0 has a length of 500. Unfortunately, this was wrong in some cases. I definitely have to use the difference between two consecutive entries for the same track, meaning that blocks here are 1000 bytes long (1000 - 0 for track 0 and 1500 - 500 for track 1). This was the bug I fixed today.

The second bug was that these MPEG blocks may contain SPU packets that actually do not belong to the current subtitle track. I forgot to throws these away. This was the bug I fixed yesterday.

You see, actually I don't know that much about the VobSub format. Maybe this is still wrong, I'm not sure. I hope not, though ;) And I'm quite hopeful as your test has not shown any more problems. BTW: A big thank you for taking the time to verify a complete movie!

Human_USB

11th January 2004, 04:35

Ok.... I thing I have found an error, but not the one I was taking about above. I'm still trying to pin down that one. Ok, on to this bug.

When I muxed 3 AC3s and taged them I get this

http://animepenguin.web1000.com/doom9/error.jpg

Here is the MMG I saved;

[mkvmergeGUI]
file_version=1
gui_version=0.8.0
output_file_name=C:\\DVD\\Rips\\SAMURAI_X_REFLECTION_VTS_02_PGC1\\1.mkv
[input]
number_of_files=5
[input/file\ 0]
file_name=C:\\DVD\\Rips\\SAMURAI_X_REFLECTION_VTS_02_PGC1\\1.avi
container=2
no_chapters=0
no_attachments=0
no_tags=0
number_of_tracks=1
[input/file\ 0/track\ 0]
type=v
id=0
enabled=1
content_type=XVID
default_track=0
aac_is_sbr=0
language=none
track_name=
cues=default
delay=
stretch=
sub_charset=default
tags=
timecodes=
display_dimensions_selected=0
aspect_ratio=16/9
display_width=
display_height=
fourcc=
compression=
[input/file\ 1]
file_name=C:\\DVD\\Rips\\SAMURAI_X_REFLECTION_VTS_02_PGC1\\1 AC3 T01 3_2ch 448Kbps DELAY -66ms.ac3
container=6
no_chapters=0
no_attachments=0
no_tags=0
number_of_tracks=1
[input/file\ 1/track\ 0]
type=a
id=0
enabled=1
content_type=AC3
default_track=1
aac_is_sbr=0
language=eng (English)
track_name=English 6 Channel
cues=default
delay=66
stretch=
sub_charset=default
tags=
timecodes=
display_dimensions_selected=0
aspect_ratio=
display_width=
display_height=
fourcc=
compression=
[input/file\ 2]
file_name=C:\\DVD\\Rips\\SAMURAI_X_REFLECTION_VTS_02_PGC1\\1 AC3 T02 2_0ch 224Kbps DELAY -66ms.ac3
container=6
no_chapters=0
no_attachments=0
no_tags=0
number_of_tracks=1
[input/file\ 2/track\ 0]
type=a
id=0
enabled=1
content_type=AC3
default_track=0
aac_is_sbr=0
language=jpn (Japanese)
track_name=Japanese 2 Channel
cues=default
delay=66
stretch=
sub_charset=default
tags=
timecodes=
display_dimensions_selected=0
aspect_ratio=
display_width=
display_height=
fourcc=
compression=
[input/file\ 3]
file_name=C:\\DVD\\Rips\\SAMURAI_X_REFLECTION_VTS_02_PGC1\\1 AC3 T03 2_0ch 224Kbps DELAY -66ms.ac3
container=6
no_chapters=0
no_attachments=0
no_tags=0
number_of_tracks=1
[input/file\ 3/track\ 0]
type=a
id=0
enabled=1
content_type=AC3
default_track=0
aac_is_sbr=0
language=eng (English)
track_name=English 2 Channel
cues=default
delay=66
stretch=
sub_charset=default
tags=
timecodes=
display_dimensions_selected=0
aspect_ratio=
display_width=
display_height=
fourcc=
compression=
[input/file\ 4]
file_name=C:\\DVD\\Projects\\SAMURAI_X_REFLECTION_VTS_02_PGC1\\SAMURAI_X_REFLECTION.idx
container=9
no_chapters=0
no_attachments=0
no_tags=0
number_of_tracks=1
[input/file\ 4/track\ 0]
type=s
id=0
enabled=1
content_type=VobSub
default_track=0
aac_is_sbr=0
language=eng
track_name=English
cues=default
delay=
stretch=
sub_charset=default
tags=
timecodes=
display_dimensions_selected=0
aspect_ratio=
display_width=
display_height=
fourcc=
compression=zlib
[attachments]
number_of_attachments=1
[attachments/attachment\ 0]
file_name=C:\\DVD\\Rips\\SAMURAI_X_REFLECTION_VTS_02_PGC1\\VTS_02 - Stream Information.txt
description=
mime_type=text/plain
style=0
[global]
segment_title=
enable_splitting=0
split_by_size=1
split_after_bytes=
split_after_time=
split_max_files=
dont_link=1
previous_segment_uid=
next_segment_uid=
chapters=C:\\DVD\\Projects\\SAMURAI_X_REFLECTION_VTS_02_PGC1\\SAMURAI_X_REFLECTION_ChapterX5.txt
chapter_language=eng (English)
chapter_charset=
cue_name_format=
global_tags=

I muxed a second video that has 3 AC3 audios just like this one but the error is not there.

-Jason

virux

11th January 2004, 04:48

isn't that the auto loading of audio files by mpc where u hav the demuxed ac3's in the same dir as the video?

Human_USB

11th January 2004, 05:07

Yes it was...... thank you. I'm to sleepy to post the other bug so I will post it tomorrow.

jk888

11th January 2004, 10:49

Originally posted by Mosu
You see, actually I don't know that much about the VobSub format. Maybe this is still wrong, I'm not sure. I hope not, though ;) And I'm quite hopeful as your test has not shown any more problems. BTW: A big thank you for taking the time to verify a complete movie! [/B]

My pleasure, I am glad to have contributed my 2 cents to this absolutely excellent app. Thx Mosu for all the work you've done.

Also, I have sent a pm to gabest (the vobsub developer) about this issue, in case he has anything to add to it.

filewalker

13th January 2004, 23:04

I have a problem with the split function.

Input is (about 1400 Mb):
1 XviD video
2 LC AAC (in MP4) files
1 .idx and .sub (with 2 subtitels inside)
1 attachment ( 1 jpeg picture)
1 ogg chapter list

In the log file stands:
Pass 1 took 254 seconds.
Pass 2: merging the files. This will take even longer.

Ok. there's no error message but after 10 minutes I quitted MKVMerge...the progress slider seemed frozen.

Have i wait longer than 10 minutes or is it a bug.
I tried it 4 times and everytime it's the same.

Here is the logfile (http://www.geocities.com/filewalker666/logfile.zip) before I quit ( copy &paste the link adress in a fresh new browser - Geocities doesn't allow direct linking)

I used the new runtime package with mkvtoolnix-0.8.1-build20040110-1

Ps.
Muxing without splitting works fine.

Cu filewalker

Mosu

13th January 2004, 23:13

Originally posted by filewalker
I have a problem with the split function.

Input is (about 1400 Mb):
1 XviD video
2 LC AAC (in MP4) files
1 .idx and .sub (with 2 subtitels inside)
1 attachment ( 1 jpeg picture)
1 ogg chapter list

In the log file stands:
Pass 1 took 254 seconds.
Pass 2: merging the files. This will take even longer.

Ok. there's no error message but after 10 minutes I quitted MKVMerge...the progress slider seemed frozen.

I can't say much about this. Please try to find out which file/track is posing the problem by eliminating one file at a time from this list and then re-muxing with splitting enabled.

Have i wait longer than 10 minutes or is it a bug.
I tried it 4 times and everytime it's the same.

Well, if the progress bar is frozen for that long then it'll definitely not continue.

Here is the logfile (http://www.geocities.com/filewalker666/logfile.zip)

Unfortunately there's nothing in it that gives me any clue.

Human_USB

13th January 2004, 23:52

I can't get the error to show back up.... so I will keep testing for it and maybe one day it will show.

-Jason

filewalker

14th January 2004, 01:41

I tested again

splitting options:
"Enable Splitting" --> ticked
"by size" --> 703 Mb
"Don't link files" --> unticked
"max.number of files" --> 2

1) video alone --> splitting was sucessful (time 3 min.) :)
2) video + 1 LC AAC(inside MP4) --> splitting was sucessful (time 6 min.) :)

3) video + 2 LC AAC files(inside MP4) --> splitting was not sucessful (I waited 23 min.) :(
(the Audio MP4 files alone play fine in Zoomplayer)

if I opened the WindowsTaskManager I saw that MKVMerge and MKVMerge GUI showed "Keine Rückmeldung" (normally is "wird ausgeführt").
It seems to be a little unstable in this situation...I wanted to save the log file but I had no access to the GUI anymore.

The prozess bar froze in the middle of this bar ( similiar to the attempt when I made the log file in the post above).

Here's the CL (http://www.geocities.com/filewalker666/CL_AVI_2AAC.zip) ( copy & paste the link adress in a fresh new browser - Geocities doesn't allow direct linking)

Hope this information helps...

Cu filewalker :)

Mosu

14th January 2004, 10:35

Originally posted by filewalker
I tested again

splitting options:
"Enable Splitting" --> ticked
"by size" --> 703 Mb
"Don't link files" --> unticked
"max.number of files" --> 2

1) video alone --> splitting was sucessful (time 3 min.) :)
2) video + 1 LC AAC(inside MP4) --> splitting was sucessful (time 6 min.) :)

3) video + 2 LC AAC files(inside MP4) --> splitting was not sucessful (I waited 23 min.) :(

That narrows it down a bit. Unfortunately I haven't been able to reproduce it, but that may be because I only have one mp4 file to begin with. Could you upload your two MP4 files to my FTP server please (see my signature)? If not I'll have to create some more MP4 files myself and I don't really have a lot of time at the moment, so I'd prefer to have your files.

The command line looks normal to me, too.

Mosu

14th January 2004, 10:35

Originally posted by Human_USB
I can't get the error to show back up.... so I will keep testing for it and maybe one day it will show.

I SO love bugs that users can't reproduce ;) Anyway, if you come across any then let me know.

filewalker

14th January 2004, 14:36

Hi mosu,

the .zip file is named "2_LC_AAC_ (inside MP4).zip".
It's about 311 Mb...so it'll last some time till it's uploaded...please be patiently.

Btw.
File linking doesn't work in Zoomplayer.
So, I'll ask Blight to implement this nice feature in Zoomplayer.
Which information(documentaion) is necessary for Blight?
(I read the the shipped MKVMerge docu, paragraph "File linking"...has it enough information ? )

Cu filewalker

Mosu

14th January 2004, 16:21

Originally posted by filewalker
Hi mosu,

the .zip file is named "2_LC_AAC_ (inside MP4).zip".
It's about 311 Mb...so it'll last some time till it's uploaded...please be patiently.

I am. Patient that is. Thanks for the upload.

Btw.
File linking doesn't work in Zoomplayer.
So, I'll ask Blight to implement this nice feature in Zoomplayer.
Which information(documentaion) is necessary for Blight?
(I read the the shipped MKVMerge docu, paragraph "File linking"...has it enough information ? )

I think it should be enough. One should maybe note that file linking has nothing to do with filenames - it's done purely with the segment IDs. Of course a player should try to find files with a similar in the same directory and then check their segment IDs when searching for the previous/next segment. That'll be faster than scanning all files in a directory. But that's purely up to the player developper - from Matroska's side there's only the segment IDs and nothing more.

karl_lillevold

14th January 2004, 22:02

I apologize if this already has been discussed, but I was experimenting with methods to extract the AAC info from RM, mainly because when transferring AAC files via RealPlayer 10 to the iPod, it transcodes to MP3 @ 96 kbps, but also just because I wanted to see if it is possible.

Test file: http://www.lillevold.com/files/aac_samples/rm_aac128.rm

I tried two methods.

1) Gabest's RealMediaSplitter -> MatroskaMuxer -> filewriter:

Creates .mka file which plays perfectly:

c:\>mkvinfo test.mka
+ EBML head
+ Segment
|+ Seek head (subentries will be skipped)
|+ EbmlVoid (size: 77)
|+ Segment information
| + Timecode scale: 1000000
| + Duration: 9.891s
| + Date: Wed Jan 14 13:25:22 2004 UTC
| + Muxing application: DirectShow Matroska Muxer
|+ Segment tracks
| + A track
| + Track number: 1
| + Track UID: 38
| + Track type: audio
| + MinCache: 1
| + MaxCache: 1
| + Codec ID: A_AAC/MPEG2/LC
| + Audio track
| + Sampling frequency: 44100.000000
| + Channels: 2
| + Bit depth: 16
|+ Cluster

Then I used mkvextract to create test.aac.
Plays fine in foobar2000.

Then I used mp4creator60 to create test.mp4
(mp4creator60 -aac-profile=4 -optimize -create=test.aac test.mp4)

Plays fine in foobar2000.
Renamed to test.m4a.
Imported to iTunes, transferred to iPod, playing fine.

2) with mkvtoolnix 0.8.1:

Then I tried mkvmerge to replace MatroskaMuxer. It does the following:

C:\>mkvmerge -o test3.mka rm_aac128.rm
Using RealMedia demultiplexer for rm_aac128.rm.
+-> Using generic audio output module for stream 0 (FourCC: raac).
Opened 'test3.mka' for writing.
progress: 118/118 packets (100%)
Muxing took 0 seconds.

The resulting file is not playable. MPC loads its RealAudio Decoder, but there is no sound.

It is probably better to identify AAC in Matroska as method 1) does, instead of just 'raac', right?

Also note that for HE-AAC both FourCC codes racp and potentially also raac can be used, leaving it to the decoder to properly identify the existence of SBR info in the case when HE-AAC is marked as raac. This seems to be a fairly common problem though, since it is my understand there is no way to flag HE in a raw .aac file.

gabest

14th January 2004, 22:20

I'm not entirely sure about this, but I think the file created by mkvmerge does not play because the new waveformat tags were not set in the saved WAVEFORMATEX struct. The rm splitter fills it with values similar to ac3 and dts, I just continued numbering them starting from 0x2002. They can be found in moreuuids.h at http://cvs.sourceforge.net/viewcvs.py/guliverkli/guliverkli/include/

(by looking at the code of the ra decoder filter, this shouldn't be a problem..., karl: do you still have that file? I'm laying on my back with a notebook on me, and a bit lazy to produce test files right now, you know... :))

karl_lillevold

14th January 2004, 22:33

All files here (https://tobor2.prognet.com/~klillevold/1056663966.59.15880/).
rm_aac128.rm
gabest.mka/aac/m4a created with MatroskaMuxer chain.
mkvmerge.mka created with mkvmerge.

Even though mkvmerge.mka could be made to play.. it would not extract correctly to AAC:

C:\>mkvextract tracks mkvmerge.mka 1:mkvmerge.aac
Warning: Unsupported CodecID 'A_REAL/RAAC' for ID 1. Skipping track.
Nothing to do. Exiting.

Mosu

14th January 2004, 22:37

Originally posted by karl_lillevold
I apologize if this already has been discussed, but I was experimenting with methods to extract the AAC info from RM, mainly because when transferring AAC files via RealPlayer 10 to the iPod, it transcodes to MP3 @ 96 kbps, but also just because I wanted to see if it is possible.

Test file: http://www.lillevold.com/files/aac_samples/rm_aac128.rm

I tried two methods.

1) Gabest's RealMediaSplitter -> MatroskaMuxer -> filewriter:

Creates .mka file which plays perfectly:

c:\>mkvinfo test.mka
+ EBML head
+ Segment
|+ Seek head (subentries will be skipped)
|+ EbmlVoid (size: 77)
|+ Segment information
| + Timecode scale: 1000000
| + Duration: 9.891s
| + Date: Wed Jan 14 13:25:22 2004 UTC
| + Muxing application: DirectShow Matroska Muxer
|+ Segment tracks
| + A track
| + Track number: 1
| + Track UID: 38
| + Track type: audio
| + MinCache: 1
| + MaxCache: 1
| + Codec ID: A_AAC/MPEG2/LC
| + Audio track
| + Sampling frequency: 44100.000000
| + Channels: 2
| + Bit depth: 16
|+ Cluster

Then I used mkvextract to create test.aac.
Plays fine in foobar2000.

Then I used mp4creator60 to create test.mp4
(mp4creator60 -aac-profile=4 -optimize -create=test.aac test.mp4)

Plays fine in foobar2000.
Renamed to test.m4a.
Imported to iTunes, transferred to iPod, playing fine.

Just to summarize it - this works perfectly?

2) with mkvtoolnix 0.8.1:

Then I tried mkvmerge to replace MatroskaMuxer. It does the following:

C:\>mkvmerge -o test3.mka rm_aac128.rm
Using RealMedia demultiplexer for rm_aac128.rm.
+-> Using generic audio output module for stream 0 (FourCC: raac).
Opened 'test3.mka' for writing.
progress: 118/118 packets (100%)
Muxing took 0 seconds.

The resulting file is not playable. MPC loads its RealAudio Decoder, but there is no sound.

It is probably better to identify AAC in Matroska as method 1) does, instead of just 'raac', right?

Definitely. Could you please give me some more info on the FourCCs that RealAudio can use for AAC?

Also note that for HE-AAC both FourCC codes racp and potentially also raac can be used, leaving it to the decoder to properly identify the existence of SBR info in the case when HE-AAC is marked as raac. This seems to be a fairly common problem though, since it is my understand there is no way to flag HE in a raw .aac file.

Yes, without starting to decode the first frame it's impossible to tell SBR and non SBR apart if the source is a .aac file. So about the FourCCs. Does racp always identify HE-AAC, and is raac potentially HE? Are there other FourCCs than these two?