EVOB De/Multiplexers

[Chumbo]

The latest BeHappy (http://forum.doom9.org/showthread.php?p=970096#post970096) may be of interest to some here.

[vathimenil]

@Pelican 9: If you're still tracking this thread... First, thanks for all the work you've put into this. Most of us wouldn't have gotten anywhere had it not been for your demux tool.

Second, I have a question: did you ever resolve your 0-byte problem when applying JnZ's method for audio decoding? I run into it as well for certain HD DVD files (e.g., full KK movie), but not others (e.g., smaller dd+ samples). If that method still isn't working for you --are you using some other method that actually works? Would love to get your insights. Thanks. Cheers.

[Pelican9]

@Pelican 9: If you're still tracking this thread... First, thanks for all the work you've put into this. Most of us wouldn't have gotten anywhere had it not been for your demux tool.

Second, I have a question: did you ever resolve your 0-byte problem when applying JnZ's method for audio decoding? I run into it as well for certain HD DVD files (e.g., full KK movie), but not others (e.g., smaller dd+ samples). If that method still isn't working for you --are you using some other method that actually works? Would love to get your insights. Thanks. Cheers.
Yes, I'm still here. :)
What is this 0-byte problem?
Have I missed something?
I'm working on SUPread now.

[Jack-Bauer]

hey guys

ive been following your thread for a while now and have been experimenting with my hd backups. you keep mentioning this sonic audio decoder 4.2. i cant find that anywhere? theres no mention of it on any site besides this one :)
if you could help me out id greatly appreciate it.


Have a good day. Jack

[Clown shoes]

hey guys

ive been following your thread for a while now and have been experimenting with my hd backups. you keep mentioning this sonic audio decoder 4.2. i cant find that anywhere? theres no mention of it on any site besides this one :)
if you could help me out id greatly appreciate it.

Sonic's audio decoder 4.2 is actually part of Sonic's Cineplayer HD DVD Decoder 4.2. The decoder that is installed with Scenarist 4.1 (standard and advanced) is Cineplayer 4.1 and it appears to be broken. Hardware support can be switched on in the options tab, but it resets itself when the options box is closed. This is fixed in 4.2 and means elementary .vc1 and .264 streams can be played back smoothly.

@Madshi, your little Eac3toAc3 tool is brilliant. I have one small suggestion though. Is it possible to have a way of selecting the ac3 bitrate? I am experimenting with the reauthoring of HD-DVDs from HD discs to DVD9s and at present the DD+ stream cannot be correctly reintegrated, therefore I am transcoding to ac3. Unfortuanately although the ac3 spec allows the use of a 640kbps stream, Scenarist HD is still limited to 448Kbps, at least in the Standard content edition. I have not had the opportunity to test Advanced content edition.

@pelican9, I beleive you may have mentioned it earlier in the thread but is there a way of accurately measuring the delay on a demuxed audio stream. I am currently trying to do it by eye in the Scenarist simulation window, which is far from ideal.

[Pelican9]

@pelican9, I beleive you may have mentioned it earlier in the thread but is there a way of accurately measuring the delay on a demuxed audio stream. I am currently trying to do it by eye in the Scenarist simulation window, which is far from ideal.

Delay between the video and audio streams?
You can see the first PTS value of every stream in EVOdemux status window, if the two value is equal than no delay.
PTS (in hexadecimal format) / 90000 = time in seconds

[Clown shoes]

OK, I think I understand. So the value that I am looking out for is the PTS. If there is a difference in that figure between the video and audio stream, I divide that figure by 90000 to get my delay in seconds?

Sorry, I'm not at my authoring machine right now so I don't have Evodemux in front of me. I hope I didn't get that too confused :)

Does anyone have any idea what is causing the problem with remuxing the dd+ audio back into an HD project. It seems to cause very bad stuttering in the video stream when muxed.

[vathimenil]

I've tried this, but it creates only a 0 byte file. :(

@Pelican 9: sorry about the delay --when running graphedit and DTS/AC3/DD+ --> Sonic Cinemaster Audio 4.2 --> Dump, your output was coming out as a 0-byte raw file. I get the same problem with some files, but not others --and I can't seem to find any rhyme or reason to which work and which do not, except maybe size of the input...

[Pelican9]

@Pelican 9: sorry about the delay --when running graphedit and DTS/AC3/DD+ --> Sonic Cinemaster Audio 4.2 --> Dump, your output was coming out as a 0-byte raw file. I get the same problem with some files, but not others --and I can't seem to find any rhyme or reason to which work and which do not, except maybe size of the input...

Thanks! I've tried one file only (UNILOGO).

[Applecore]

thanks a lot! :)

ive been trying to dump the .raw file now using graphedit but when i try to connect the dd+ file with the sonic audio decoder 4.2 i get an error message saying that they cannot agree on a connection.

[Rectal Prolapse]

Delay between the video and audio streams?
You can see the first PTS value of every stream in EVOdemux status window, if the two value is equal than no delay.
PTS (in hexadecimal format) / 90000 = time in seconds

You're my hero Pelican! On one of my rips I had to estimate a delay of 750 ms - when I look at the PTS information of the tracks, the difference s 667 ms, which is close!

DELETED: Nevermind - I was looking at the wrong VC1 stream.

Thanks!

[madshi]

Here's version 1.1 of the tool "eac3to" (the tool formerly known as "eac3toac3"):

http://madshi.net/eac3to.zip

Changes:
(1) Support for FLAC encoding added.
(2) Bitrate can be specified via command line parameter.
(3) ffdshow removed from the filter chain.
(4) "ddp" and "ec3" file extensions are accepted now, too.
(4) Fix: "dd+" file extension didn't work correctly.

The reason why I had ffdshow in the chain was that during early experimentation I often only had 2 valid channels in the converted wav files and 4 trash channels. But after some more testing it seems that this problem only occurs when using the Wav Dest filter. When directly connecting the Sonic Audio Decoder to the Dump filter the problem doesn't seem to occur.

FLAC support is complete and works fine. Unfortunately, the current flac.exe (version 1.1.4) still has a limit of 2GB for raw input files. So we'll have to wait for a newer flac.exe to make things work properly.

[Rectal Prolapse]

I am having issues with WavPack playback - if I encode the WavPack stream as 24 bit, I get A/V sync problems - it gets worse the longer I seek into the movie!

However, it works fine in MPC - I don't know why it goes out of sync when using GraphEdit or TheaterTek (with custom graph).

This only seems to occur with 24 bit encoded WavPack streams. Any ideas why it only works in MPC, even though the filters are the same? (I disabled the internal MPC filters - so using the same 1.7.00.18 Haali Media Splitter filter).

EDIT: I found a workaround - if I add FFDShow's Audio Processor after the output of the WavPack Decoder, the problem goes away. I figured this out after attaching GraphEdit to MPC, and found that MPC adds the Audio Switcher filter to the WavPack Decoder outputs - so maybe it somehow kept everything in sync. Seems to work with FFDShow Audio Processor too! I am happy! :D

I still don't understand how MediaCoder overcomes the FLAC 2 GB limitation - it uses the FLAC commandline tool as far as I can tell, using the Process Explorer program to check the command line as it runs.

[Rectal Prolapse]

@Pelican 9: sorry about the delay --when running graphedit and DTS/AC3/DD+ --> Sonic Cinemaster Audio 4.2 --> Dump, your output was coming out as a 0-byte raw file. I get the same problem with some files, but not others --and I can't seem to find any rhyme or reason to which work and which do not, except maybe size of the input...

Have you tried moving the position slider and pressing play? Also, hit play, then move the slider, then push play again. This is the only way I found to get dump to give me anything other than a zero byte file.

I am using the DD+ soundtrack from the Harry Potter HD-DVD.

EDIT: Put in the edit in the wrong post! oops! :blush:

[Rectal Prolapse]

I still don't understand how MediaCoder overcomes the FLAC 2 GB limitation - it uses the FLAC commandline tool as far as I can tell, using the Process Explorer program to check the command line as it runs.

I just had a thought - I believe FLAC can handle the INPUT of 2 GB of WAV bytes. However, it is likely that it chokes when it tries to OUTPUT more than 2 GB of FLAC. I will try to test this.

[Jan2001]

Here's version 1.1 of the tool "eac3to" (the tool formerly known as "eac3toac3"):

http://madshi.net/eac3to.zip

Changes:
(1) Support for FLAC encoding added.
(2) Bitrate can be specified via command line parameter.
(3) ffdshow removed from the filter chain.
(4) "ddp" and "ec3" file extensions are accepted now, too.
(4) Fix: "dd+" file extension didn't work correctly.

The reason why I had ffdshow in the chain was that during early experimentation I often only had 2 valid channels in the converted wav files and 4 trash channels. But after some more testing it seems that this problem only occurs when using the Wav Dest filter. When directly connecting the Sonic Audio Decoder to the Dump filter the problem doesn't seem to occur.

FLAC support is complete and works fine. Unfortunately, the current flac.exe (version 1.1.4) still has a limit of 2GB for raw input files. So we'll have to wait for a newer flac.exe to make things work properly.

Thanks a lot for that tool.
First time I got a proper 6 channel ac3 file and no crap anymore.:thanks:

[vathimenil]

Have you tried moving the position slider and pressing play? Also, hit play, then move the slider, then push play again. This is the only way I found to get dump to give me anything other than a zero byte file.

I am using the DD+ soundtrack from the Harry Potter HD-DVD.

EDIT: I found a workaround - if I add FFDShow's Audio Processor after the output of the WavPack Decoder, the problem goes away. I figured this out after attaching GraphEdit to MPC, and found that MPC adds the Audio Switcher filter to the WavPack Decoder outputs - so maybe it somehow kept everything in sync. Seems to work with FFDShow Audio Processor too! I am happy! :D
Wow... not quite sure how that works (shaking my head in disbelief), but it does. Go figure... but THANKS!!!

[Rectal Prolapse]

Ignore the EDIT part of that post vathimelli - it was supposed to be in my previous posted - but - it might just work too...hmmm. Nope - adding FFDShow Audio Processor before dump doesn't work.

[Darth Pinous]

I just had a thought - I believe FLAC can handle the INPUT of 2 GB of WAV bytes. However, it is likely that it chokes when it tries to OUTPUT more than 2 GB of FLAC. I will try to test this.
Confirmed. I tried a FLAC reencode of a 3,7 GB WAV file. If I use the same parameters as Mediacoder, the encode starts but it stops as soon as the FLAC file reaches 2 GB.

[JnZ]

Confirmed. I tried a FLAC reencode of a 3,7 GB WAV file. If I use the same parameters as Mediacoder, the encode starts but it stops as soon as the FLAC file reaches 2 GB.
Try rewrite WAVe using Wavewizard...it can fix broken header in WAVes over 2 or 4GB.

[Darth Pinous]

Try rewrite WAVe using Wavewizard...it can fix broken header in WAVes over 2 or 4GB.

The input WAV file has already been rewritten by Wavewizard, problem here is FLAC that doesn't want to go over 2 GB for the output file.

[MichalHabart]

I a having problems with Rundown HDDVD. When i demux all tracks from evo files, i am not able able to load mpa audio tracks into graphedit using orbitlee's filter (i always receive error that filter can not load data). I suspect evodemux made something wrong because at the end, i did not get any summary like is normal (i mean counted frames and so on). Anyone else had similar problems?

@Pelican, orbitlee: Is there something i am doing wrong?

Edit: I have evodemux version 0.623b

Edit2: According to evodemux, it should be DD+ at 1536kbps with length something about 1 hour. but when i try to play it thru MPC (with orbitlee's filter), i get lenght more then 6 hours and instead of playing, it just counts errors :(

[zgx]

I think MediaCoder tricks FLAC.exe so it does not know how big the file is. You can try this just for fun (it works but you get a small click the first second).

flac.exe -o output.flac - < largefile.raw

(you probably also need to add stuff like --endian, --channels, --bps, --sample-rate, --signed)

[madshi]

I think MediaCoder tricks FLAC.exe so it does not know how big the file is. You can try this just for fun (it works but you get a small click the first second).

flac.exe -o output.flac - < largefile.raw

(you probably also need to add stuff like --endian, --channels, --bps, --sample-rate, --signed)
That was a good hint!

Here's version 1.2 of the "eac3to" tool:

http://madshi.net/eac3to.zip

Changes:
- FLAC encoding works now without any input/output size limits

Disadvantage: The only way to make FLAC work with >2GB input and output files results in FLAC files which don't have a run length in the header. So if you e.g. play the resulting FLAC file in foobar2000, foobar2000 will not know how long the file plays and you won't be able to seek in the file. However, muxing the FLAC track into a MKV movie works just great and then seeking and everything works fine again.

I've made some size comparisons and here's the result:

Serenity English track, 1:59:02, 48khz, 24bit, 5.1 channels

raw/wav : 5,74 GB (100%)
flac : 2,66 GB ( 46%)
e-ac3 1536kbps : 1,27 GB ( 22%)
ac3 640kbps : 0,53 MB ( 9%)
So the 20-30% compression number for FLAC that was given earlier in this thread seems to be wrong. I think the 20-30% number comes from reducing the bitdepth from 24bit to 16bit plus FLAC compression. If I do that, I end up with 30% file size.

[SBeaver]

That was a good hint!

Serenity English track, 1:59:02, 48khz, 24bit, 5.1 channels

raw/wav : 5,74 GB (100%)
flac : 2,66 GB ( 46%)
e-ac3 1536kbps : 1,27 GB ( 22%)
ac3 640kbps : 0,53 MB ( 9%)
So the 20-30% compression number for FLAC that was given earlier in this thread seems to be wrong. I think the 20-30% number comes from reducing the bitdepth from 24bit to 16bit plus FLAC compression. If I do that, I end up with 30% file size.

What would you say is an apropriate bitrate for ogg 5.1 with audio like that?
It is a matter of opinion surely but what would be the equivalent sound quality of a 16-bit DTS track like those on a DVD today?

[orbitlee]

I a having problems with Rundown HDDVD. When i demux all tracks from evo files, i am not able able to load mpa audio tracks into graphedit using orbitlee's filter (i always receive error that filter can not load data). I suspect evodemux made something wrong because at the end, i did not get any summary like is normal (i mean counted frames and so on). Anyone else had similar problems?

@Pelican, orbitlee: Is there something i am doing wrong?

Edit: I have evodemux version 0.623b

Edit2: According to evodemux, it should be DD+ at 1536kbps with length something about 1 hour. but when i try to play it thru MPC (with orbitlee's filter), i get lenght more then 6 hours and instead of playing, it just counts errors :(



A few megabytes sample please. You can cut with any hex editor.

[MichalHabart]

A few megabytes sample please. You can cut with any hex editor.

Here is link to 20MB of mpa file from evodemux:
http://www.sendspace.com/file/uojxqv

If it is not going to be enough, i will demux whole track and cut more MBs from it.

[Applecore]

how did you guys remux an encoded mkv file with a dd+ audiotrack? mkvmerge seems to have a problem with my track.
is that even possible?

[zgx]

how did you guys remux an encoded mkv file with a dd+ audiotrack? mkvmerge seems to have a problem with my track.
is that even possible?Not sure what version of mkvmerge that added dd+ support but if it doesn't work with the latest official release you can check out the latest pre build at http://www.bunkus.org/videotools/mkvtoolnix/win32/pre/.

[zgx]

That was a good hint!

So the 20-30% compression number for FLAC that was given earlier in this thread seems to be wrong. I think the 20-30% number comes from reducing the bitdepth from 24bit to 16bit plus FLAC compression. If I do that, I end up with 30% file size.I'm glad I could be of help :). When I did it myself I got a clicking sound the first second did that happen to you?

I think many movie mixes doesn't use the rear channels and LFE that much and thats great for FLAC compression. But if you have audio all the time on all channels you won't get that much compression. If you do reduce bitdepth to 16 bit (if the source was 24) you will of course get even smaller files. I encoded the audio track for Old Boy and the RAW 24 bit file was huge (6-7GB or something) and the 16 bit FLAC I got was under 1GB.

[madshi]

I'm glad I could be of help :). When I did it myself I got a clicking sound the first second did that happen to you?
No, not at all. Actually I did this test:

(1) Encode short raw file with flac.exe the usual way.
(2) Encode short raw file with flac.exe by using MediaEncoders command line trick.
(3) Decode both flac files to wav.
(4) Bitwise compare of the wav files.

They final wav files were 100% identical. So when using "eac3to" + "flac" you'll get no clicking sound - unless it's in the original source. Don't know about MediaEncoder. Maybe there's a bug in MediaEncoder that causes the clicking sound, I've no idea. E.g. I could imagine that if you treat a WAV header as raw PCM data, you could get a clicking sound. But "eac3to" really feeds raw data to FLAC and no WAV header. So no danger there...

I encoded the audio track for Old Boy and the RAW 24 bit file was huge (6-7GB or something) and the 16 bit FLAC I got was under 1GB.
I'm wondering how big/small the 24->16 bit quality loss is. I'm also wondering whether the original E-AC3 track has real 24 bits to begin with. The sampling rate is listed in the E-AC3 header, but the bit depth doesn't seem to be. Also I believe to have read that Aften converts everything to 32 bit floating point before encoding. So I've really no idea to which bit depth an (E-)AC3 decoder should really decode to.

[orbitlee]

Here is link to 20MB of mpa file from evodemux:
http://www.sendspace.com/file/uojxqv

If it is not going to be enough, i will demux whole track and cut more MBs from it.


I checked the sample, looks like there are some garbage before the first audio frame. You should use hex editor to remove all bytes before the blue block (start with 0B 77)


http://img136.imageshack.us/img136/7701/sampledq9.th.jpg (http://img136.imageshack.us/my.php?image=sampledq9.jpg)

@Pelican9, Is it possible that EVO contains partial audio frame in the first 2048 bytes pack?

[Clown shoes]

I'm trying to fix the delay on my ec3 file but I'm not used to working with hexidecimal figures. Can anyone tell me what I need to do.

The first PTS of my VC1 stream is 00000D75

The first PTS of my EC3 stream is 000008FE

Thanks

[Pelican9]

I'm trying to fix the delay on my ec3 file but I'm not used to working with hexidecimal figures. Can anyone tell me what I need to do.

The first PTS of my VC1 stream is 00000D75

The first PTS of my EC3 stream is 000008FE

Thanks

D75=3445
EC3=3779
diff=334 -> 3,7 ms

Powered by windows calculator...

[Pelican9]

Here is link to 20MB of mpa file from evodemux:
http://www.sendspace.com/file/uojxqv

If it is not going to be enough, i will demux whole track and cut more MBs from it.

Could you share the first 20-50 MB of your EVO file?

[MichalHabart]

I checked the sample, looks like there are some garbage before the first audio frame. You should use hex editor to remove all bytes before the blue block (start with 0B 77)


http://img136.imageshack.us/img136/7701/sampledq9.th.jpg (http://img136.imageshack.us/my.php?image=sampledq9.jpg)

@Pelican9, Is it possible that EVO contains partial audio frame in the first 2048 bytes pack?

Well, check with evodemux will find DD+ track before it finds video track. could it be that audio starts before video and it is causing this problem?

[Clown shoes]

Powered by windows calculator...

lol :p Things you never knew windows calculator could do

3.7 ms! that is very weird. If a frame is aprox 41 ms then 3.7 will be imperceptable, whereas my audio is falling a number of seconds late. Is there anything else that might be causing such a delay?

[MichalHabart]

Could you share the first 20-50 MB of your EVO file?

How can i cut it, with hex editor too?

And do you know any powerfull and free hex editor that can handle files that big?

[orbitlee]

How can i cut it, with hex editor too?

And do you know any powerfull and free hex editor that can handle files that big?

I remember that drmpeg posted a tool named "head" in this thread, which could cut the head of a file, long long ago. :cool:

[zgx]

3.7 ms! that is very weird. If a frame is aprox 41 ms then 3.7 will be imperceptable, whereas my audio is falling a number of seconds late. Is there anything else that might be causing such a delay?I have similar problems. I can't get my DD+ audio and VC-1 video to sync. Have tried both to mux it into MKV and to play the EVO directly. The audio is totally out of sync and after a few seconds the audio just studders.

Below is the graph I use when I try to play the EVO:

http://img368.imageshack.us/img368/5703/notworkingky7.jpg

Anyone know what I am doing wrong?

[Deckard2019]

The first PTS of my VC1 stream is 00000D75
The first PTS of my EC3 stream is 000008FE

D75=3445
EC3=3779
diff=334 -> 3,7 ms

It's not EC3 but 8FE :D
So :
D75=3445
8FE=2307
diff=1138 -> 12,64 ms

[MichalHabart]

Could you share the first 20-50 MB of your EVO file?

Here is 30MBs of evo file:
http://rapidshare.com/files/21179483/cut.evo.html

[Clown shoes]

Well spotted Deckard, but I take it we are still talking milliseconds and not seconds, so nothing that would explain a delay of about 10 seconds.

Now I think about it, I wonder if the logo at the front of the feature is anywhere near 10 seconds?

[Rectal Prolapse]

zgx, what splitter are you using? Is it the EAC3 (DD+) track you muxed in?

Weird - I can never get Sonic Cinemaster Audio Decoder 4.2 to connect with Haali - it never agrees to connect! Any ideas on why that is? Oh - the decoder doesn't work on Dolby TrueHD streams, even when using Sonic HD Demuxer. I wonder if that is related?

[MichalHabart]

Well spotted Deckard, but I take it we are still talking milliseconds and not seconds, so nothing that would explain a delay of about 10 seconds.

Now I think about it, I wonder if the logo at the front of the feature is anywhere near 10 seconds?

I had similar problem and it seemed to be caused by x264 and different framerate. When i use xvid in avi, both tracks (dd+ and video) have framerate 23.976. but i tried x264 in mkv (created by megui) and video was 23.976 but audio was for strange reason 24fps. This led to big async during the movie. Try change framerate of video to 24fps if it will be better.

I assume it is problem of mkv merge program.

[Pelican9]

It's not EC3 but 8FE :D
So :
D75=3445
8FE=2307
diff=1138 -> 12,64 ms

Yessss!
:thanks:

[MichalHabart]

zgx, what splitter are you using? Is it the EAC3 (DD+) track you muxed in?

Weird - I can never get Sonic Cinemaster Audio Decoder 4.2 to connect with Haali - it never agrees to connect! Any ideas on why that is? Oh - the decoder doesn't work on Dolby TrueHD streams, even when using Sonic HD Demuxer. I wonder if that is related?

At the moment Sonic Audio decoder can not play TrueHD tracks, we have to wait for newer version.

[zgx]

zgx, what splitter are you using? Is it the EAC3 (DD+) track you muxed in?

Weird - I can never get Sonic Cinemaster Audio Decoder 4.2 to connect with Haali - it never agrees to connect! Any ideas on why that is? Oh - the decoder doesn't work on Dolby TrueHD streams, even when using Sonic HD Demuxer. I wonder if that is related?I'm using the latest Haali Media Splitter to open the EVO and yes it's a DD+ track in the EVO file. The EVO was created by using EVOdemux to merge the two feature EVO files and also remove all audio tracks but the DD+ track I wanted.

I have no idea why it doesn't work for you. It works for me for DD+ tracks (except for the sync problem) but it don't think it works for TrueHD yet.

[Pelican9]

Here is 30MBs of evo file:
http://rapidshare.com/files/21179483/cut.evo.html

What a mess!
Is this an original movie?
I guess your file is broken because it contains very strange things...
Missing headers, packets, lots of zero byte fields, etc....

[madshi]

I'm using the latest Haali Media Splitter to open the EVO and yes it's a DD+ track in the EVO file. The EVO was created by using EVOdemux to merge the two feature EVO files and also remove all audio tracks but the DD+ track I wanted.

I have no idea why it doesn't work for you. It works for me for DD+ tracks (except for the sync problem) but it don't think it works for TrueHD yet.
For me the best solution is to use Haali Media Splitter + Haali Matroska Muxer. The resulting MKV plays fine for me. If that doesn't work for you, try only muxing the VC-1 into MKV first. In the next step demux the DD+ track and then mux it into the VC-1 MKV by using mkvtoolnix. That should work fine, too.