mkvtoolnix (Matroska toolkit): new release

[karl_lillevold]

@Mosu: Yes, method 1) works perfectly, the iPod plays the result fine.
The FourCC codes are raac and racp.
raac can be both AAC and HE-AAC
racp should be only HE-AAC.
Our AAC encoder, licensed from Coding Technologies, is LC or LC+SBR. The decoder can handle everything. I am not an expert in the various flavors of AAC, but I have been told, MPEG-2 and MPEG-4 are identical when neither SBR nor PNS are used.
See this folder (http://www.lillevold.com/files/aac_samples/) for the samples I have, AAC, HE-AAC, 2ch and 5.1. All with 4cc raac though.

[Mosu]

Thanks Karl, that's all the info I needed for now. I'm already implementing mapping to the proper CodecID (and then mkvextract should work on the result, too).

[gabest]

Mosu: I mean this should be done when muxing aac from rm to matroska. edit: oh, you are right, there is nothing, but in the media header of rm there is exactly what we need to reconstruct.

[Mosu]

Originally posted by gabest
Mosu: I mean this should be done when muxing aac from rm to matroska. edit: oh, you are right, there is nothing, but in the media header of rm there is exactly what we need to reconstruct.

Do you mean the type_specific_data ? If yes... where exactly? In the sample file it is 81 bytes long, and that's even more than the 'normal' real_audio_v5_properties structure that I know of...

[gabest]

Just go the end of the v5 struct, add 4 (only 3 for v4) and you will find a DWORD (extra length) plus the same ammount of bytes till the end to type_specific_data. These bytes seem to have a 0x02 at the beginning always, but the rest is what we are looking for. We could put this into codec private and the matroska splitter wouldn't have to reconstruct it, just when it cannot find anything there.

edit:
One more thing. In case you wanted to demux the aac frames already from the rm packets. The first WORD is the number of packets in this form: 00 x0 (you can ignore it if you want), then the frame sizes each a WORD too, and finally the frames themselves till the end of the packet. The time length of a frame can be calculated knowing that it has 2048 samples.

[Mosu]

Hmm Karl, this could be problematic. As far as I can see you store AAC headerless (which is good, normally), but one realaudio packet contains more than one AAC frame. Example from your rm_he-aac96_51.rm:


| + Block (track number 1, 8 frame(s), timecode 0.000s)
| + Frame with size 1117
| + Frame with size 1117
| + Frame with size 1117
| + Frame with size 1116
| + Frame with size 1117
| + Frame with size 1116
| + Frame with size 1117
| + Frame with size 1116
| + Block group
| + Block (track number 1, 8 frame(s), timecode 0.743s)


At 44100Hz sample rate this amounts to 44100 * 0.743 / 8 = 4096 samples per packet, and AAC normally only has 1024 samples per packet. Can you shed some light on this, please? 'Cause for Matroska's native modes it's imperative that there's only one media frame per Matroska blobk.

[Mosu]

Originally posted by gabest
Just go the end of the v5 struct, add 4 (only 3 for v4) and you will find a DWORD (extra length) plus the same ammount of bytes till the end to type_specific_data. These bytes seem to have a 0x02 at the beginning always, but the rest is what we are looking for. We could put this into codec private and the matroska splitter wouldn't have to reconstruct it, just when it cannot find anything there.

I don't want to put anything into CodecPrivate for AAC because we don't do that for the other sources (.AAC / .MP4 files) either - and there should be consistency.

edit:
One more thing. In case you wanted to demux the aac frames already from the rm packets. The first WORD is the number of packets in this form: 00 x0 (you can ignore it if you want), then the frame sizes each a WORD too, and finally the frames themselves till the end of the packet. The time length of a frame can be calculated knowing that it has 2048 samples.

Gabest, you're my hero :) You've just answered the question I've just posted a minute ago :)

2048 samples per packet? Phew... so far I haven't seen any AAC file with != 1024 samples per packet. Interesting.

[gabest]

Mosu: I'm sure it's 2048 samples. Did you consider the channel number too? In case of a stereo file, there can be 1024 samples per channel, but this gets even more lower for 6 channels. All samples I tested so far prooved this. About putting anything into codecprivate. Since we are already reconstructing it, why not make it more perfect if we can? The most I can reconstruct is 2 bytes now, that means we loose 3 from the end. Don't know if they ever could be useful, but I'd prefer to keep them.

edit: Just verified rm_he-aac96_51.rm, it also has 2048 samples, one frame is 464399 long measured in dshow's 100ns units. In the mentioned example there are two aac frames inside one rm packet and since there are 8 packets the total time is 2*8*464399 = 7430384 ~ 0.743 s.

edit2: Something strange is going on, now I found another file and it confuses me :P

edit3: Don't know what to think now. If 1024 sample per channel is true, then the 6 channel files also have 2 channels or we should calculate with that at least. Or not. Don't believe anything I say at 1:07am.

[Human_USB]

LOL.... my the way I got the error muxing 5 vobsubs in one idx with 2 AC3s and one video.

[Stux]

1024 samples per long block,

thus 1024 * channels per frame

except for SBR there are 2048 samples per block.

The 3ivx audio decoder requires 1 frame per media sample

[Mosu]

Originally posted by Stux
1024 samples per long block,

thus 1024 * channels per frame

except for SBR there are 2048 samples per block.

The 3ivx audio decoder requires 1 frame per media sample

To be honest I don't get it. You surely can't say that SBR-AAC is always 2048 samples per AAC frame because that's simply not true. I have plenty of SBR AAC files that have 1024 samples per AAC frame, and only now I've come upon this. Is there a way to extract the number of samples in a frame from it without having to decode it completely? If not we might have a problem (maybe not only we but other containers as well, dunno).

[Mosu]

Originally posted by gabest
Mosu: I'm sure it's 2048 samples. Did you consider the channel number too? In case of a stereo file, there can be 1024 samples per channel, but this gets even more lower for 6 channels. All samples I tested so far prooved this. About putting anything into codecprivate. Since we are already reconstructing it, why not make it more perfect if we can? The most I can reconstruct is 2 bytes now, that means we loose 3 from the end. Don't know if they ever could be useful, but I'd prefer to keep them.

Well... I thought that this is how to recreate the last three bytes for HE-AAC:

#define AAC_SYNC_EXTENSION_TYPE 0x02b7
codecdata[2] = AAC_SYNC_EXTENSION_TYPE >> 3;
codecdata[3] = ((AAC_SYNC_EXTENSION_TYPE & 0x07) << 5) | 5;
codecdata[4] = (1 << 7) | (output_samplerate_idx << 3);


with output_samplerate_idx being the index for the output samplerate.

edit: Just verified rm_he-aac96_51.rm, it also has 2048 samples, one frame is 464399 long measured in dshow's 100ns units. In the mentioned example there are two aac frames inside one rm packet and since there are 8 packets the total time is 2*8*464399 = 7430384 ~ 0.743 s.

Yes, that's the same file I'm testing at the moment.

edit3: Don't know what to think now. If 1024 sample per channel is true, then the 6 channel files also have 2 channels or we should calculate with that at least. Or not. Don't believe anything I say at 1:07am.

It's definitely not 1024 samples per channel per frame. I've been muxing AAC (HE and none HE) for a couple of months now, and until now every file had 1024 samples per frame although they were mostly stereo. A lot of others have used mkvmerge with other AAC files as well, and always 1024 per frame was used...

[filewalker]

Hi mosu,

Thanks for the hint with the "resume" function... :)

The .zip file is now complete.
Both MP4 files contain LC AAC, VBR/stereo, 100-120 Kbps, made with Nero/latest BeSweet


Hope it helps,


Cu filewalker

[Stux]

Originally posted by Mosu
To be honest I don't get it. You surely can't say that SBR-AAC is always 2048 samples per AAC frame because that's simply not true. I have plenty of SBR AAC files that have 1024 samples per AAC frame, and only now I've come upon this. Is there a way to extract the number of samples in a frame from it without having to decode it completely? If not we might have a problem (maybe not only we but other containers as well, dunno).

An HE-AAC stream will have 2048 * channels samples per frame
An LC-AAC stream will have 1024 * channels samples per frame

(unless short block coding is used... )

The thing is, an HE-AAC stream has a base layer LC-AAC stream, which *does* have 1024*Channels samples...

(perhaps my definition of "frame" is different to yours :-\)

[spyder]

Do you mean they decode to 2048 samples since it is actually half samplerate or is it really 2048 samples coded in a frame?

EDIT: Disregard my question. I apparently am too sleepy to have understood the discussion until I read it again ;)

[Mosu]

Originally posted by Stux
An HE-AAC stream will have 2048 * channels samples per frame
An LC-AAC stream will have 1024 * channels samples per frame

(unless short block coding is used... )

The thing is, an HE-AAC stream has a base layer LC-AAC stream, which *does* have 1024*Channels samples...

(perhaps my definition of "frame" is different to yours :-\)

I still don't fully understand it, but that's most likely because I don't know much about AAC in general to begin with. So far I've chosen to just calculate the number of samples from the timestamps of the RealMedia file and not rely on any other assumption.

[KpeX]

Hi Mosu,

I have a question regarding the new tagging system. I've been tagging some MKA audio files using the Foobar2000 plugin (due to its freeDB capabilities).

However, if I try to extract these tags with MKVtoolnix 0.8.1 (mkvextract.exe tags "filename.mka"), I get no output at all. Additionally, if I run a 'mkvinfo -v' on the files, the tags applied by foobar are recognized correctly and displayed at the end of the output. Thanks again,

[Mosu]

Originally posted by KpeX
Hi Mosu,

I have a question regarding the new tagging system. I've been tagging some MKA audio files using the Foobar2000 plugin (due to its freeDB capabilities).

However, if I try to extract these tags with MKVtoolnix 0.8.1 (mkvextract.exe tags "filename.mka"), I get no output at all. Additionally, if I run a 'mkvinfo -v' on the files, the tags applied by foobar are recognized correctly and displayed at the end of the output. Thanks again,

Can you please upload such a tagged file to my FTP server? (see my sig)

Thanks.

[KpeX]

Originally posted by Mosu
Can you please upload such a tagged file to my FTP server? In progress now. Thanks :).

[Mosu]

Originally posted by KpeX
In progress now. Thanks :).

That file is broken. The problem is that yes, the tags are present at the end of the file, but no, they are not part of the segment. Therefore none of my tools will 'see' those tags. And no again, this is not a valid Matroska file.

You should bug the author of the software you used.

[KpeX]

Unfortunately jccston has discontinued the plugin (http://forum.doom9.org/showthread.php?s=&threadid=67850). Hopefully the fb2k developer will continue it, I'll post this on HA.

Thanks for your work, Mosu, it is greatly appreciated.

[Mosu]

Heya,

here's another release of mkvtoolnix, 0.8.2. A few usability enhancements together with a couple of minor bug fixes make up this release. Please note that Windows users need to re-download the runtime package also available on my home page because I've switched to more DLLs.

The mkvtoolnix homepage:
http://www.bunkus.org/videotools/mkvtoolnix/
The sources:
http://www.bunkus.org/videotools/mkvtoolnix/sources/mkvtoolnix-0.8.2.tar.bz2
The Windows binaries and the Windows runtime package:
http://www.bunkus.org/videotools/mkvtoolnix/win32/mkvtoolnix-0.8.2.rar
http://www.bunkus.org/videotools/mkvtoolnix/win32/mkvtoolnix-runtime.rar

Here's the ChangeLog since 0.8.1:
-----------------------
2004-01-21 Moritz Bunkus <moritz@bunkus.org>
* Released v0.8.2.

2004-01-19 Moritz Bunkus <moritz@bunkus.org>
* mkvmerge: bug fix: The PCM handling was broken resulting in packets that did not end on sample boundaries.

2004-01-17 Moritz Bunkus <moritz@bunkus.org>
* mkvmerge: bug fix: AVIs with uncompressed sound were leading to buffer overflows.
* mkvmerge/mmg: allow the track names to be empty so that you can remove them when muxing Matroska files. Same for the file title.
* mkvmerge: new feature: The track headers will be rendered completely including the elements that are set to their default values. Causes less confusion and allows the setting of e.g. the track language without having to remux the file completely.
* mkvmerge: bug fix: If remuxing a file that contains frames with a reference to the same timecode those references were lost turning such P frames into I frames. This was the case for some RealAudio stuff.

2004-01-15 Moritz Bunkus <moritz@bunkus.org>
* mmg: new feature: Automatically pre-set the attachment's MIME type if the file has a known extension (e.g. 'text/plain' for '.txt').
* mkvmerge: new feature: Unknown/unsupported track types can be copied 1:1 from Matroska input files.
* mkvmerge: new feature: Added proper support for AAC-inside-RealMedia files.

2004-01-14 Moritz Bunkus <moritz@bunkus.org>
* mkvmerge: new feature: Write cues for audio-only files as well (not more than one cue entry during a two seconds period).

2004-01-12 Moritz Bunkus <moritz@bunkus.org>
* mkvmerge: bug fix: The default track flags could not be overriden on the command line when reading Matroska files.
* Windows binaries after v0.8.1 require a new runtime DLL archive. Please download it from http://www.bunkus.org/videotools/mkvtoolnix/ Thanks.

2004-01-11 Moritz Bunkus <moritz@bunkus.org>
* mkvmerge: new feature: Added the two new chapter flags 'hidden' and 'enabled'.
* mkvmerge: new feature: Added a new format for the external timecode files.

2004-01-09 Moritz Bunkus <moritz@bunkus.org>
* mkvmerge: bug fix: The VobSub handling was on occasion putting SPU packets for the wrong MPEG stream into the current stream resulting in that particular entity not being displayed.
-----------------------

Have fun :)

[outlyer]

Originally posted by Mosu
here's another release of mkvtoolnix, 0.8.2.Cool, as always :)

BTW, a semi-bug :p, the configure script checks for libebml 0.6.2, while apparently 0.6.4 is required, at least the 0.6.3 I had installed failed at compilation time.

[karl_lillevold]

mkvmerge: new feature: Added proper support for AAC-inside-RealMedia files.
excellent!

Seems to work fine for AAC in RM, but not HE-AAC in RM, or at least MPC 6.4.7.5 can not play back the resulting HE-AAC files, either with 3ivx or CoreAAC. Yes I used the --aac-is-sbr flag. Files in this folder (http://www.lillevold.com/files/aac_samples/). mkvmerge still works fine for ADTS .aac with HE-AAC.


mkvinfo rm_he-aac32.mka
+ EBML head
+ Segment, size 45833
|+ Seek head (subentries will be skipped)
|+ EbmlVoid (size: 4031)
|+ Segment information
| + Muxing application: libebml v0.6.4 + libmatroska v0.6.3
| + Writing application: mkvmerge v0.8.2
| + Duration: 9.984s
| + Date: Wed Jan 21 06:25:05 2004 UTC
| + Segment UID: 0xa1 0xac 0x9b 0xf1 0x0b 0x60 0x16
0x81 0x03 0x7b 0xcf 0x0a 0xe2 0xb2 0xda 0x2c
|+ Segment tracks
| + A track
| + Track number: 1
| + Track UID: 2952442546
| + Track type: audio
| + MinCache: 0
| + Timecode scale: 1.000000
| + Codec ID: A_AAC/MPEG4/LC/SBR
| + Default flag: 1
| + Language: eng
| + Audio track
| + Sampling frequency: 44100.000000
| + Channels: 2
|+ EbmlVoid (size: 1024)

I notice that Codec ID says MPEG4, vs when crated from ADTS, it says MPEG2. This may not matter. However, then I notice that for the RM file, the sampling frequency says 44100. I think this is not right. It should probably say 22050, with output sampling frequency 44100, no?

ADTS AAC -> MKA mkvinfo:

+ Codec ID: A_AAC/MPEG2/LC/SBR
+ Audio track
+ Sampling frequency: 22050.000000
+ Channels: 2
+ Output sampling frequency: 44100.000000


Oh, and one more little thing: in 'mmg' the checkbox for "AAC is SBR/HE-AAC/AAC+" is grayed out when opening RM file...

[Mosu]

Originally posted by karl_lillevold
excellent!

Seems to work fine for AAC in RM, but not HE-AAC in RM, or at least MPC 6.4.7.5 can not play back the resulting HE-AAC files, either with 3ivx or CoreAAC. Yes I used the --aac-is-sbr flag. Files in this folder (http://www.lillevold.com/files/aac_samples/). mkvmerge still works fine for ADTS .aac with HE-AAC.

I notice that Codec ID says MPEG4, vs when crated from ADTS, it says MPEG2. This may not matter. However, then I notice that for the RM file, the sampling frequency says 44100. I think this is not right. It should probably say 22050, with output sampling frequency 44100, no?

The MPEG4 vs MPEG2 should not matter, but yes, the rest was a bug. I think I've fixed it, and I can now play the two non-HE samples and rm_he-aac32 fine, but for rm_he-aac96_51 libfaad cannot decode any frame, so I guess that (at least for me) the file was muxed fine, but I cannot verify it. Please try this build: http://www.bunkus.org/videotools/mkvtoolnix/win32/pre/mkvtoolnix-0.8.2-build20040121-1.rar

Oh, and one more little thing: in 'mmg' the checkbox for "AAC is SBR/HE-AAC/AAC+" is grayed out when opening RM file...

Thanks, this should of course not happen. (a bit later) It does not happen for me - I can select it just fine for all the four sample files on your server. But with the new code you shouldn't have to check it anyway because mkvmerge parses the AAC config data that is stored in the RealMedia audio headers and gets its settings from there.

[karl_lillevold]

Mosu: your new build works without any problems for HE-AAC in RM to HE-AAC in MKA. I can play back all the resulting HE-AAC files, incl rm_he-aac96_51, the latest CoreAAC and 3ivx AAC decoders both handle it. Thanks!

I still can not check the box for SBR in mmg, but that no longer matters. Oh, that reminds me, next time you update your list of filetypes in mmg, could add a few more to RealMedia : (.ra, .rv, .ram? ra is RealAudio. rv will be RealVideo, ram is the result of web server filename remapping (RM files are saved as .ram when served via http, at least from apache).

[Mosu]

Originally posted by karl_lillevold
Mosu: your new build works without any problems for HE-AAC in RM to HE-AAC in MKA. I can play back all the resulting HE-AAC files, incl rm_he-aac96_51, the latest CoreAAC and 3ivx AAC decoders both handle it. Thanks!

Great, thanks for the feedback. So the playback problem I had with the rm_he-aac96_51 is probably just a result of an outdated libfaad.

I still can not check the box for SBR in mmg, but that no longer matters.

Ok, I'll check that on Windows next time I boot into it. The problem is that I've set up vmware for compilation at the moment, and while that does work fine, the resulting binaries all segfault at the end, and I have no idea why. Interesting enough they don't if used on a 'normal' PC/Windows installation.

Oh, that reminds me, next time you update your list of filetypes in mmg, could add a few more to RealMedia : (.ra, .rv, .ram? ra is RealAudio. rv will be RealVideo, ram is the result of web server filename remapping (RM files are saved as .ram when served via http, at least from apache).

Sure, I'll do that right away.

[hubereevez]

Both wmp and mpc crash when playing rv9+raac64+srt muxed by 0.8.2.....
Muxing with 0.8.1 ends with no-sound movie
++
hubhub
edit : playable with a new (browsed version): mkvtoolnix-0.8.2-build20040121-1.rar, but video is accelerating, audio not synch, mkv time 03:09 stead of 01:35

[Mosu]

Originally posted by hubereevez
edit : playable with a new (browsed version): mkvtoolnix-0.8.2-build20040121-1.rar, but video is accelerating, audio not synch, mkv time 03:09 stead of 01:35

Can you please upload that file to my FTP server (see my sig)? Obviously my method for detecting/calculating the number of samples in one AAC packet is wrong in this case.

Thanks.

[Mosu]

Originally posted by karl_lillevold
I still can not check the box for SBR in mmg

I've just checked - with all the four .rm files you have on your website I can select 'AAC is SBR...' in mmg, so I'm a bit clueless about what's going wrong on your end.

[wiflye81]

I've just try tu mux a nero mp4 with mmg and it don't work ( 'mkvmerge -i' failed. Return code : 1. Errno 0 (No Error), Make sure that you've selected a mkvmerge executable on the settings tab ).
All other files I try to mux works. Something strange, mkvmerge ( under dos ) just mux audio track, it says unknown fourCC "mp4v".
I search on the net but I can't find any mp4 fourcc changer.

[Mosu]

Originally posted by wiflye81
I've just try tu mux a nero mp4 with mmg and it don't work ( 'mkvmerge -i' failed. Return code : 1. Errno 0 (No Error), Make sure that you've selected a mkvmerge executable on the settings tab ).
All other files I try to mux works. Something strange, mkvmerge ( under dos ) just mux audio track, it says unknown fourCC "mp4v".
I search on the net but I can't find any mp4 fourcc changer.

Reading video from MP4 files is not supported at the moment. That's intentional. This is due to the topic of 'native MPEG4', see http://forum.doom9.org/showthread.php?s=&threadid=68420

On the other hand the file should still be addable to mmg... Could you please run the following on your file:

mkvmerge --identify-verbose yourfile.mp4 > output.txt

Then send me that output.txt (either by Email or post it here, it shouldn't be longer than two or three lines anyway).

Thanks.

[filewalker]

Hi mosu, :)

I tested the new version, but I still have the problem with muxing the 2 AAC files (which are uploaded).

What i noticed is that an icon popped up in the system tray, which says that I have too "less virtuell memory".

So I set manually the virtaull memory to a higher level.
But there is no difference if I started MKVMerge again.

It stops mostly at the same position...(see the progressbar in the sreenshot), the created MKV file is mostly about 625 Mb

Here's a screenshot from the "Windows task manager"
http://www.geocities.com/filewalker666/MKVMerge.gif(copy & paste in a fresh new browser window)

MKVMerge GUI shows "Keine Rückmeldung"

Maybe there's a memory administration problem with MKVMerge...

Cu filewalker :)

[Mosu]

Originally posted by filewalker
I tested the new version, but I still have the problem with muxing the 2 AAC files (which are uploaded).

Ah yes, sorry, I had totally forgotten about this. I'll get right to it.

[Mosu]

Originally posted by filewalker
Hi mosu, :)

A stupid memleak. Please try this build: http://www.bunkus.org/videotools/mkvtoolnix/win32/pre/mkvtoolnix-0.8.2-build20040122-1.rar

[karl_lillevold]

Originally posted by Mosu
I've just checked - with all the four .rm files you have on your website I can select 'AAC is SBR...' in mmg, so I'm a bit clueless about what's going wrong on your end.
Sorry, silly user error on my end. When I click on the RM file I just added in the input file box, a new world of options open up. Before this everything was grayed out. I even knew this from when I used mmg to mux multiple RM audio files and had to set different audio languages... Very nice tool to handle the multitude of mkvmerge options!

[Mosu]

Originally posted by karl_lillevold
Sorry, silly user error on my end. When I click on the RM file I just added in the input file box, a new world of options open up. Before this everything was grayed out.

Yes, the number of options available to the user depends on the currently selected track. If neither a file nor a track has been selected yet everything is grayed out.

[The Shemeta]

Hi,

I am trying to install Mkvtoolnix under Linux & i always get the same error. so i thought i can ask you guys for help. here is what i get:

depmode=gcc3 /bin/sh ../../depcomp \
g++ -DHAVE_CONFIG_H -I. -I. -I../.. -I../.. -I../../src -I../../avilib-0.6.10 -I../../aviclasses -I. -Wall -Wno-sign-compare -Wno-comment -D_LARGEFILE64_SOURCE -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -Wshadow -c -o chapters.o `test -f 'chapters.cpp' || echo './'`chapters.cpp
chapters.cpp: In function `libebml::EbmlMaster*
copy_chapters_recursive(libebml::EbmlMaster*)':
chapters.cpp:585: `KaxChapterFlagHidden' undeclared (first use this function)
chapters.cpp:585: (Each undeclared identifier is reported only once for each
function it appears in.)
chapters.cpp:585: syntax error before `::' token
chapters.cpp:586: `esrc' undeclared (first use this function)
chapters.cpp:586: `edst' undeclared (first use this function)
chapters.cpp:587: syntax error before `;' token
chapters.cpp:588: syntax error before `*' token
chapters.cpp:592: `KaxChapterFlagEnabled' undeclared (first use this function)
chapters.cpp:592: syntax error before `::' token
chapters.cpp:594: syntax error before `;' token
chapters.cpp:595: syntax error before `*' token
make[3]: *** [chapters.o] Error 1
make[3]: Leaving directory `/home/stanislav/downloads/mkvtoolnix-0.8.2/src/common'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/home/stanislav/downloads/mkvtoolnix-0.8.2/src'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/stanislav/downloads/mkvtoolnix-0.8.2'
make: *** [all] Error 2

i hope you can help me solve my problem.

thanks in advance!

[Mosu]

Originally posted by The Shemeta
Hi,

I am trying to install Mkvtoolnix under Linux & i always get the same error. so i thought i can ask you guys for help. here is what i get:

Update your libebml and libmatroska from www.matroska.org

[The Shemeta]

are there any new releases of matroska? that's what i have installed

checking FLAC/format.h presence... yes
checking for FLAC/format.h... yes
checking for EBML... yes
checking Ebml version >= 0.6.2... ok (0.6.3)
checking for Matroska... yes
checking Matroska version >= 0.6.2... ok (0.6.2)
checking for XML_ParserCreate in -lexpat... yes
checking expat.h usability... yes
checking expat.h presence... yes

[Mosu]

Originally posted by The Shemeta
are there any new releases of matroska? that's what i have installed

checking FLAC/format.h presence... yes
checking for FLAC/format.h... yes
checking for EBML... yes
checking Ebml version >= 0.6.2... ok (0.6.3)
checking for Matroska... yes
checking Matroska version >= 0.6.2... ok (0.6.2)
checking for XML_ParserCreate in -lexpat... yes
checking expat.h usability... yes
checking expat.h presence... yes

Yes: http://matroska.free.fr/downloads/libebml/libebml-0.6.4.tar.gz and http://matroska.free.fr/downloads/libmatroska/libmatroska-0.6.3.tar.gz

I forgot to update the configure checks, though - it should abort on anything older than these two.

[Rash]

Just a question Mosu. Is that splitting feature keyframe accurate? And (I obviously know the answer, but I don't think it will harm to ask :)) can I split VobSubs with it?

Thanks a lot. ;)

[Mosu]

Originally posted by Rash
Just a question Mosu. Is that splitting feature keyframe accurate? And (I obviously know the answer, but I don't think it will harm to ask :)) can I split VobSubs with it?

Thanks a lot. ;)

Yes, the split function will always produce files that are a) smaller than the split argument (be it size or time) and b) end right before a key frame. The following files all start with a key frame.

Note that mkvmerge wil not handle files with multiple video streams gracefully if splitting is activated! Only the first video stream will be taken into account.

Yes, splitting VobSubs with it is no problem whatsoever.

[tiki4]

@Mosu:

Sorry,

I've got a somewhat related questions: I compiled the new libebml and libmatroska in order to compile mkvtoolnix-0.8.2. However, I then tried to recompile MPlayer-1.0pre3 against the new libs, but compilation didn't run through because of some undeclared functions. It compiles fine, if I step down to 0.6.3 and 0.6.2. Should I upgrade to MPlayer CVS or is there another problem? I think the ChangeLog of libmatroska mentioned some clean-up...

Thanks,

tiki4

P.S.: SuSE 9.0 with gcc-3.3.1

[Mosu]

Originally posted by tiki4
@Mosu:

Sorry,

I've got a somewhat related questions: I compiled the new libebml and libmatroska in order to compile mkvtoolnix-0.8.2. However, I then tried to recompile MPlayer-1.0pre3 against the new libs, but compilation didn't run through because of some undeclared functions. It compiles fine, if I step down to 0.6.3 and 0.6.2. Should I upgrade to MPlayer CVS or is there another problem? I think the ChangeLog of libmatroska mentioned some clean-up...

Thanks,

tiki4

P.S.: SuSE 9.0 with gcc-3.3.1

We're currently in the process of fixing the Matroska 1.0 specs. We've therefore deactivated some elements that were not really used. Unfortunately the Matroska demuxer in 1.0pre3 still contains some of those elements. You should upgrade to the CVS version. It contains a new Matroska demuxer written in C that doesn't even require libebml and libmatroska anymore.

[wiflye81]

here is the output.txt :
Warning: Quicktime/MP4 reader: Unknown/unsupported FourCC 'mp4v' for track 3.
File 'title.mp4': container: Quicktime/MP4
Track ID 3: video (mp4v)
Track ID 4: audio (mp4a)

[The Shemeta]

Hi,

i installed mkvtoolnix yesterday & i read on the homepage that it had a GUI too. but the comman "mmg" doesn't work on my comp. i am using linux. did i do something wrong or do i have to ./configure with extra oarameters for the gui?

Thanks for the help,

see ya around!

[Mosu]

Originally posted by The Shemeta
Hi,

i installed mkvtoolnix yesterday & i read on the homepage that it had a GUI too. but the comman "mmg" doesn't work on my comp. i am using linux. did i do something wrong or do i have to ./configure with extra oarameters for the gui?

The GUI uses wxWindows 2.4 or newer. You'll have to install the appropriate development packages as well. On Debian they're called libwxgtk2.4 and libwxgtk2.4-dev. configure will try to find those libs and print out something like:

checking for wxWindows... 2.4.2 ok

[tiki4]

Thanks. Will do so.

tiki4

[The Shemeta]

Thanks for the info, it was very usefull. i just did my first mkv file with AAC sound, which plays great on Mplayer.

great work Mosu! Danke shoen! ;)