eac3to - audio conversion tool

[baudi]

This is a Haali splitter bug, also present in Unforgiven HD DVD, with VC-1 codec. Fortunately only a few VC-1 movies have this problem with Haali splitter.

If you mux the VC-1 stream into a .ts container you will see a perfect video, no rainbow images.

[intomed]

Can someone point me in the right direction as to how to accomplish this? I apologize for my ignorance, it's just that this is my first journey into video conversion.

[rickardk]

This is a Haali splitter bug, also present in Unforgiven HD DVD, with VC-1 codec. Fortunately only a few VC-1 movies have this problem with Haali splitter.

If you mux the VC-1 stream into a .ts container you will see a perfect video, no rainbow images.

Hmm.. Strange because I have remuxed Unforgiven and it plays without problems. Blu-ray source but I think they share the same VC-1 encode.

[Wackyphill]

Yes. The truehd track doesn't need any delay, so eac3to didn't apply any to the flac. But it should be in sync.

Is the original evo file in sync?
If yes, do you notice the same de-sync throughout the remuxed (mkv) movie? Try using different audio/video decoders to see if the de-sync still exists.

Well, I don't have software for playing the evo so I can't say but I think it would be fine, it's just a straight rip from anydvd.

The audio seems to be off by a constant amount I don't think it drifts.

Is the fact their seems to be 2 video streams in the evo possibly the problem?

I'm using ffmpeg to play the FLAC. What other decoder would you suggest?

[Chumbo]

@madshi,
Would you consider adding an option to not apply delay automatically please? I noticed upon demuxing an e-ac3 track, that the delay was applied but I couldn't turn it off. I tried -0ms and +0ms and neither worked.

The delay is actually valid in that the audio doesn't start until several seconds into the short clip I have. A good example of this is the Paramount HD logo EVO on the Hunt for Red October HD DVD or probably most other Paramount HD DVDs. :)

[Joniii]

Is TrueHD -> FLAC completely lossless or is there some process while converting that doesn't make it possible?

[napalm-187]

hi i decoding true hd track from constantin

nero shows 24 bit

eac3to writing 16 bit and 24 waves, after transcoding, the 24 bit files deleted automaticly,
here is the log txt

i think true hd have always 24 bit?


eac3to v2.14
"D:\Downloads\eac3\eac3to\eac3to.exe" "F:\test.thd" "F:\test.wavs" -nero
TrueHD, 5.1 channels, 48khz, dialnorm: -27dB
Removing dialog normalization...
Decoding with DirectShow (Nero Audio Decoder 2)...
DirectShow reports 5.1 channels, 24 bits, 48khz
Writing WAVs...
Creating/writing file "F:\test.24bit.R.wav"...
Creating/writing file "F:\test.24bit.L.wav"...
Creating/writing file "F:\test.24bit.C.wav"...
Creating/writing file "F:\test.24bit.SL.wav"...
Creating/writing file "F:\test.24bit.SR.wav"...
Creating/writing file "F:\test.24bit.LFE.wav"...
This audio track contains only 16 bit of information.
The zero bytes were successfully removed.
eac3to processing took 10 minutes, 14 seconds.
Done.

[madshi]

However
eac3to.exe HS.dtshd HS.dts
should either reencode to dts or do nothing an throw an error. just copying the file seems strange.
It does not just copy the file, it also removes dialnorm (if used) and it patches the stream to 24bit (if it's 16bit and not Master Audio). Finally, it removes zero padding (if contained in the stream).

Anyway, the reason why "eac3to source.dtshd dest.dts" is not extracting the core, anymore, is that eac3to v2.0 does not even look at the source file extension, anymore. Instead eac3to is looking inside the file to find out what audio stream type it is. So you could name the source file "source.funny" and eac3to would still work just fine. And "dts" is a valid file extension for both DTS and DTS-HD. So using "dts" as the destination file extension doesn't necessarily mean that you want to have the DTS-HD blocks stripped.

eac3to v1.x was not able to do any processing on DTS-HD streams except extracting the core. E.g. just removing dialnorm or zero padding from a DTS-HD stream was simply not supported. Because of that eac3to v1.x always automatically extracted the core if the source file was DTS-HD and the destination file was DTS. Since eac3to v2.x can do many things with DTS-HD and core extraction is only one of those things, core extraction is now only done if you ask for it.

My Surcode installation is fine. When I use the dts core and try to reencode everything is fine. Reencoding the dtshd file to flac doesn't work either.
In that case seemingly the Sonic decoder stalls when decoding your DTS-HD file. It could be a bug in the Sonic decoder. Or it could be some kind of corruption in the DTS-HD blocks. DTS is usually not protected by a CRC, so there's no way to check if the data is corrupt or not. I've already noticed that decoding can stall or crash if there are bit faults somewhere in the stream.

If you believe that this might be a bug in eac3to then I'd suggest that you try GraphEdit (File Async Source -> Sonic Demuxer -> Sonic Audio Decoder -> Dump) to see whether that graph runs through. If it does, eac3to is likely to have a bug. If GraphEdit stalls, too, it's either a bug in the Sonic Audio Decoder or a corruption in the DTS-HD track.

[nautilus7]

Well, I don't have software for playing the evo so I can't say but I think it would be fine, it's just a straight rip from anydvd.

The audio seems to be off by a constant amount I don't think it drifts.

Is the fact their seems to be 2 video streams in the evo possibly the problem?

I'm using ffmpeg to play the FLAC. What other decoder would you suggest?
Having 2 evo files wouldn't be a problem. Try madshi's madflac decoder. It can be found in this forum.

[madshi]

F:\HDDVD\300\demux\eac3to\eac3to.exe F:\HDDVD\300\demux\feature_300NDOM6LF1VC1_HD1.EVO+F:\HDDVD\300\demux\feature_300NDOM6LF1VC1_HD1_Divide.EVO 5: F:\HDDVD\300\demux\1.flac
Is that the USA HD DVD of 300? I'm a bit confused by the names of the EVO files. This "blabla_devide.evo" names are usually only used by European studios (e.g. Studio Canal) but not by US studios!

Anyway, is the audio out of sync in the beginning of the movie already? In that case it'd be very helpful if you could upload a sample of the movie for me to test. Don't know how long the sample has to be. Depends on how soon the audio sync problem is noticable.

Also, could you please check whether the E-AC3 tracks show the same audio sync problems? Thanks.

[madshi]

However, I have been unable to get a clean video demux from the POTO HD DVD. I have ripped and re-ripped this movie with no success. The audio comes out wonderfully. However, the video is severely unwatchable with, for lack of a better term, rainbow blocks.
As was already posted by Snowknight26 and baudi this is a known problem with POTO. I have the same problem when trying to play the EVO files directly in MPC with the Haali Media Splitter. Currently I don't know how to solve this.

[madshi]

Is the fact their seems to be 2 video streams in the evo possibly the problem?
Nope, that's no problem.

I'm using ffmpeg to play the FLAC. What other decoder would you suggest?
Try madFlac, but I rather doubt that it will make a difference with audio sync. But it'd be worth a try...

[madshi]

@madshi,
Would you consider adding an option to not apply delay automatically please? I noticed upon demuxing an e-ac3 track, that the delay was applied but I couldn't turn it off. I tried -0ms and +0ms and neither worked.

The delay is actually valid in that the audio doesn't start until several seconds into the short clip I have. A good example of this is the Paramount HD logo EVO on the Hunt for Red October HD DVD or probably most other Paramount HD DVDs. :)
You can do that right now. If eac3to e.g. lists that 100ms delay is necessary it will automatically apply this delay. Now if you don't like this, just do "-100ms"... :)

Just for my interest: For what purpose do you need that?

[madshi]

Is TrueHD -> FLAC completely lossless or is there some process while converting that doesn't make it possible?
It's completely lossless, as long as there's no bug anywhere.

[madshi]

hi i decoding true hd track from constantin

nero shows 24 bit

eac3to writing 16 bit and 24 waves, after transcoding, the 24 bit files deleted automaticly,
here is the log txt

i think true hd have always 24 bit?
The decoder output is always 24bit. But the lower 8bit of the decoder output is often just filled with zeroes. eac3to detects that and then converts the FLAC file to 16bit. Don't worry, eac3to will never throw away any audio data when doing TrueHD -> FLAC, unless you explicitely tell it to.

[napalm-187]

hi
thanks for this, i think there was an error :)

dawn of the dead uncut true hd tracks goings completed ouput from 24 bit and hears damn good





The decoder output is always 24bit. But the lower 8bit of the decoder output is often just filled with zeroes. eac3to detects that and then converts the FLAC file to 16bit. Don't worry, eac3to will never throw away any audio data when doing TrueHD -> FLAC, unless you explicitely tell it to.

[madshi]

hi
thanks for this, i think there was an error :)

dawn of the dead uncut true hd tracks goings completed ouput from 24 bit and hears damn good
I'm confused. Can you please explain what you mean? What does "goings completed ouput" mean? And what error are you talking about?

[derekz]

is possible to create an installation pack with all necessary/optional software for eac3to ? (like eac3togui,haali media splitter, ecc..)

thanks

[napalm-187]

@madhsi

i mean this

the true hd track from dawn of the dead was decoding to 24 bit waves.

nero say 24 bit and nero was decoded to 24 bit,

then i decoded the true hd track from constantine, there was decoded to 16 bit, and i mean this is an error, well i think all true hd tracks have 24 bit in it,

but its not so, true hd track have booth 16 and 24 bit in it right?

[nautilus7]

Truehd is always 24 bit, BUT it is possible not all bit to be filled with info. The remaining bits (until the 24th) are filled with zeros. These zeros are removed by eac3to.

[napalm-187]

hmmm

ok and what can i do the i became 24 bit waves and no 16 bit waves after transcoding?

[madshi]

hmmm

ok and what can i do the i became 24 bit waves and no 16 bit waves after transcoding?
You just need to stop worrying... :)

Fact 1: All TrueHD tracks decode to 24 bit.
Fact 2: Some TrueHD tracks contain full 24 bit of information, others have 8 bit of those 24 bit filled up with nothing but zeroes.

eac3to detects both situations and behaves accordingly.

[napalm-187]

You just need to stop worrying... :)

Fact 1: All TrueHD tracks decode to 24 bit.
Fact 2: Some TrueHD tracks contain full 24 bit of information, others have 8 bit of those 24 bit filled up with nothing but zeroes.

eac3to detects both situations and behaves accordingly.


ahh ok this is the point,
then is for me all claer

thx madshi for your great tool

[madshi]

is possible to create an installation pack with all necessary/optional software for eac3to ? (like eac3togui,haali media splitter, ecc..)
Of course I could include them in the download. But I don't really like the idea. At least Haali's Filters and mkvtoolnix are full blown installers which create start menu items etc. eac3to in comparison is just a simple command line tool which doesn't even have an installer. So all I could do is include the Haali and mkvtoolnix installers in the eac3to.zip. You'd still have to install those software packages manually. I don't really see the big advantage of doing that.

[Joniii]

Anyone have any idea how AC3Filter handles 5.1 FLAC that has been converted from TrueHD? I mean that AC3Filter picks 5.1 on FLAC and passes it trought SPDIF into receiver, It uses it's AC3encoder with bitrate 640kbps, so my receiver says DD 640kbps. Do I lose some quality there or is it just lossless passtrough?

[nautilus7]

Like you said it is converted to DD 640 kbps. Is DD same to flac? I don't think so...

[Wackyphill]

Is that the USA HD DVD of 300? I'm a bit confused by the names of the EVO files. This "blabla_devide.evo" names are usually only used by European studios (e.g. Studio Canal) but not by US studios!

Anyway, is the audio out of sync in the beginning of the movie already? In that case it'd be very helpful if you could upload a sample of the movie for me to test. Don't know how long the sample has to be. Depends on how soon the audio sync problem is noticable.

Also, could you please check whether the E-AC3 tracks show the same audio sync problems? Thanks.



I'm almost positive it's the US version but I'll double check tonight. I live in the US and bought it from Amazon... but I'll check. Any obvious label on the case to check for? What I found unusual about it was not that it had 2 evos w/ wierd names but that there were 2 video streams in the evo w/ different framerates?

I installed your flac decoder originally but I wasn't getting FLAC playback in my player. It wasn't until I installed ffmpeg that it started playing back. How can I specify which decoder gets used?

I'll try the e-ac3. I'll have to try and force libav cuz I dont have the other decoders.

Also (because I don't want to put anyone on a wild goose chace), is it possible a slower computer could cause this problem? My PC is a little under powered for playback but I didn't think that video/audio would look and sound smooth but be behind audio I thought they'd maybe stutter together. Is that a possibility though?

[madshi]

I'm almost positive it's the US version but I'll double check tonight. I live in the US and bought it from Amazon...
Anyone else has remuxed 300 US HD DVD yet? Did you have audio sync problems, too?

What I found unusual about it was not that it had 2 evos w/ wierd names but that there were 2 video streams in the evo w/ different framerates?
That's normal. The 2nd video is PIP.

I installed your flac decoder originally but I wasn't getting FLAC playback in my player. It wasn't until I installed ffmpeg that it started playing back. How can I specify which decoder gets used?
Depends on the media player. In MPC right click on video -> Filters.

Also (because I don't want to put anyone on a wild goose chace), is it possible a slower computer could cause this problem? My PC is a little under powered for playback but I didn't think that video/audio would look and sound smooth but be behind audio I thought they'd maybe stutter together. Is that a possibility though?
If playback is fluid I'd say that the slow computer shouldn't be guilty. Furthermore if the slow computer would cause the problem the audio sync would probably be different everytime you restart the movie.

[Chumbo]

You can do that right now. If eac3to e.g. lists that 100ms delay is necessary it will automatically apply this delay. Now if you don't like this, just do "-100ms"... :)

Just for my interest: For what purpose do you need that?
Well, you're right I could do that by countering the automatic delay applied, but then that would impact the other audio tracks I may be demuxing like TrueHD. It would be easier to just have an option to not apply the delay automatically on all tracks being demuxed.

I guess I'm from the mindset that if a tool applies anything automatically, it should allow the disabling of that feature, i.e., demux as-is. :)

It's not a huge deal because I can always get the track via graphedt and Dump as-is.

[nautilus7]

I don't think you are right...

I am under the impression that you can specify different options for each output track:

eac3to first.evo+second.evo 1: video.mkv 2: audio1.ac3 -100ms 3: audio3.flac +100ms

[Chumbo]

I don't think you are right...

I am under the impression that you can specify different options for each output track:

eac3to first.evo+second.evo 1: video.mkv 2: audio1.ac3 -100ms 3: audio3.flac +100ms
Oh wow, I'm happy to be wrong if that's the case. :) Madshi, you little sneak, is this true? ;) I will try this.

[baudi]

Hmm.. Strange because I have remuxed Unforgiven and it plays without problems. Blu-ray source but I think they share the same VC-1 encode.

You can check it:

http://forum.doom9.org/showthread.php?p=1033199#post1033199

[baudi]

Can someone point me in the right direction as to how to accomplish this? I apologize for my ignorance, it's just that this is my first journey into video conversion.

Demux the .EVO file with EVODemux, then you get a .mpv file containing the VC-1 stream. Mux that .mpv to .ts with Elecard XMuxer Pro 2.5.

Good luck.

[madshi]

Oh wow, I'm happy to be wrong if that's the case. :) Madshi, you little sneak, is this true? ;) I will try this.
nautilus7 is right. Every "action" has its own set of options. E.g. "eac3to source.evo 2: audio1.ac3 -100ms 2: audio2.ac3 2: audio3.ac3 +100ms" will create 3 different files with 3 different runtimes. This also means that if you e.g. want to delay all audio tracks by 100ms, you HAVE to add "+100ms" after each and every destination file.

I guess I'm from the mindset that if a tool applies anything automatically, it should allow the disabling of that feature, i.e., demux as-is.
Ok, but what is the practical purpose of disabling "automatic delay correction"? Why would you ever want to do that?

[madshi]

Demux the .EVO file with EVODemux, then you get a .mpv file containing the VC-1 stream. Mux that .mpv to .ts with Elecard XMuxer Pro 2.5.
You can also demux the video track with eac3to, btw.

[Chumbo]

nautilus7 is right. Every "action" has its own set of options. E.g. "eac3to source.evo 2: audio1.ac3 -100ms 2: audio2.ac3 2: audio3.ac3 +100ms" will create 3 different files with 3 different runtimes. This also means that if you e.g. want to delay all audio tracks by 100ms, you HAVE to add "+100ms" after each and every destination file.


Ok, but what is the practical purpose of disabling "automatic delay correction"? Why would you ever want to do that?
Very cool madshi. What a great feature to allow that per action. :)

As to the practical purpose, I guess it's more for convenience for those who would like to use eac3to exclusively w/out having to resort to graphedt or anything else for things eac3to does already and does very well. Demuxing the elementary streams as-is I think should be essential if that makes sense.

Btw, now that you've so turbo charged eac3to, do you think it's time to rename the tool? :) btw, when do you sleep? :p

[madshi]

As to the practical purpose, I guess it's more for convenience for those who would like to use eac3to exclusively w/out having to resort to graphedt or anything else for things eac3to does already and does very well. Demuxing the elementary streams as-is I think should be essential if that makes sense.
What is the practical purpose of demuxing elementary streams as-is?

Or in other words: Why do you want to end up with audio tracks which are not in sync with the video? :confused:

Btw, now that you've so turbo charged eac3to, do you think it's time to rename the tool? :)
Do you have a new name in mind which is so great that it's worth the change? I didn't really find such a name yet... Sure, "eac3to" doesn't really fit, anymore, but this name has the advantage of being quite well known already. So a new name must be really good to be worth the change...

btw, when do you sleep? :p
At night? :p

[nesNYC]

Converting the VC1's to AVI solves a lot of the EVO playback issues I've encountered. Seems the VC1's tell direct show not to use the Sonic demux and instead uses the Windows DMO. Also, blocking the Windows video DMO in MPC does the trick also.

As was already posted by Snowknight26 and baudi this is a known problem with POTO. I have the same problem when trying to play the EVO files directly in MPC with the Haali Media Splitter. Currently I don't know how to solve this.

[Chumbo]

What is the practical purpose of demuxing elementary streams as-is?

Or in other words: Why do you want to end up with audio tracks which are not in sync with the video? :confused:

Like I mentioned in a prior post, the Paramount logo, for example, shows a delay of 918ms, however the audio becomes out of sync AFTER the delay is automatically applied. So it's not always true that automatically applying the delay "fixes" the audio so it's in sync with the video.

I think any of the Paramount HD DVDs will have PARALOGO.EVO if you want to see what I'm talking about. :)

[Snowknight26]

Converting the VC1's to AVI solves a lot of the EVO playback issues I've encountered. Seems the VC1's tell direct show not to use the Sonic demux and instead uses the Windows DMO. Also, blocking the Windows video DMO in MPC does the trick also.

Any way you look at it, muxing the VC-1 stream into an mkv will result in rainbow frames. Decoder should have influence on how its displayed (*).


* There are exceptions, but thats not the point.

[madshi]

Like I mentioned in a prior post, the Paramount logo, for example, shows a delay of 918ms, however the audio becomes out of sync AFTER the delay is automatically applied.
I didn't understand that until now. Hmmmm... I think I only have Transformers from Paramount, nothing else. Does that problem occur with the Paramount logo from Transformers, too, although it's rather unusual (h264 instead of VC-1)?

[nautilus7]

Chumbo is right. We have an issue here:

This is transformers 1st evo.

eac3to FEATURE_1.EVO
EVO/VOB, 2 video tracks, 5 audio tracks, 1:11:54
1: h264/AVC
2: VC-1, -84ms
3: E-AC3, 5.1 channels, 1536kbit/s, 48khz, dialnorm: -27dB, -84ms
4: E-AC3, 5.1 channels, 768kbit/s, 48khz, dialnorm: -27dB, -84ms
5: E-AC3, 5.1 channels, 768kbit/s, 48khz, dialnorm: -27dB, -84ms
6: E-AC3, 2.0 channels, 192kbit/s, 48khz, dialnorm: -27dB, -84ms
7: E-AC3, 2.0 channels, 192kbit/s, 48khz, dialnorm: -27dB, -111ms

[madshi]

Chumbo is right. We have an issue here:

This is transformers 1st evo.

eac3to FEATURE_1.EVO
EVO/VOB, 2 video tracks, 5 audio tracks, 1:11:54
1: h264/AVC
2: VC-1, -84ms
3: E-AC3, 5.1 channels, 1536kbit/s, 48khz, dialnorm: -27dB, -84ms
4: E-AC3, 5.1 channels, 768kbit/s, 48khz, dialnorm: -27dB, -84ms
5: E-AC3, 5.1 channels, 768kbit/s, 48khz, dialnorm: -27dB, -84ms
6: E-AC3, 2.0 channels, 192kbit/s, 48khz, dialnorm: -27dB, -84ms
7: E-AC3, 2.0 channels, 192kbit/s, 48khz, dialnorm: -27dB, -111ms
What issue do we have?

BTW, why are there no details for the video tracks? Are you sure you're using the latest eac3to build?

[nautilus7]

I mean that there are 2 video streams but 1 audio for both. What will happen when remux the main video to mkv? Maybe you can invent something to automate the process. :D That's all.

And yes, it's v2.10, cause transformers hd dvd is not at the main pc. :p

[madshi]

I mean that there are 2 video streams but 1 audio for both.
That's the case with (I don't know) 50% of the movies? Why should that be a problem?

What will happen when remux the main video to mkv?
Nothing special. The main video track will be remuxed. The audio tracks will be synced to the main video track. The 2nd video track will be ignored. That already works just fine, tested in on several movies...

[intomed]

I just finished with The Transformers. Audio and video are in sync. Very nice. Thanks Madshi.

[nautilus7]

All right then... I guess i was confused.

[Chumbo]

I didn't understand that until now. Hmmmm... I think I only have Transformers from Paramount, nothing else. Does that problem occur with the Paramount logo from Transformers, too, although it's rather unusual (h264 instead of VC-1)?
I don't have Transformers, but I kept the one from The Hunt for Red October which is avc and ddp. It's ~23MB. I can upload it somewhere if you'd like. Let me know.

[madshi]

I don't have Transformers, but I kept the one from The Hunt for Red October which is avc and ddp. It's ~23MB. I can upload it somewhere if you'd like. Let me know.
Yeah, if there's an audio sync problem with this EVO, please upload it. BTW, is there no audio sync problem when playing this EVO directly with PowerDVD?

[Chumbo]

Yeah, if there's an audio sync problem with this EVO, please upload it. BTW, is there no audio sync problem when playing this EVO directly with PowerDVD?
I actually can't play HD material via PowerDVD because my video card is incompatible. I can play avc stuff smoothly in graphedt or MPC and I don't remember any sync issues playing it directly in the original evo. I'll check tonight and upload it for you.