eac3to - audio conversion tool

[madshi]

Originally Posted by nautilus7

The processing completed fine, but the "decoding truehd track" and "removing dialnorm" was displayed twice, i guess one for each output track (flac and ac3).
Do these actions actually happen twice?

Yes...

[nautilus7]

Well... is this necessary?


Oh, i forgot to mention it before. I used nero decoder, cause the libav produced the "famous" error again. As you said earlier, it must be the only flaw to the code.

[madshi]

Originally Posted by nautilus7

Well... is this necessary?

Yes and no. Of course an intelligent human can easily see that double decoding is kind of superfluous. But trying to duplicate this with a computer program is not always that straightforward. Sure I could do it, but it would make the code a lot more complicated. So I implemented a simpler solution which sometimes results in double decoding. On the positive side, due to the simpler code there's less potential for bugs.

[nautilus7]

Ok, i understand.

Is there anything (except delay ) that would you like me to test?

[Thunderbolt8]

can you tell in how far codes or routines from other programs (for example evodemux or h264tsto or haali filters) were used regarding .evo joining & de- and remuxing? would be useful in case we get errors, so we might know which program(s) would produce a similar error in that case and dont have to try each other tool then.

[Chumbo]

Another one to report. I ran the -test switch to see it in action and note the results for the Nero Audio Decoder. Please let me know what other info I can provide:

eac3tov2 feature_1.evo+feature_2.evo -test

Nero Audio Decoder (Nero 7 or older) is not working correctly
Sonic Audio Decoder (4.3.0.169) works fine
Haali Media Splitter (2007-11-18) is installed
Surcode DTS Encoder (1.0.23.0) is installed 
MkvToolnix (v2.1.0) is installed
EVO/VOB, 2 video tracks, 4 audio tracks, 1:29:31
1: Joined EVO/VOB file
2: VC-1
3: VC-1
4: E-AC3, 5.1 channels, 384kbit/s, 48khz, dialnorm: -27dB, -8ms
5: TrueHD, 5.1 channels, 48khz, dialnorm: -24dB, -1ms
6: E-AC3, 2.0 channels, 192kbit/s, 48khz, dialnorm: -27dB, -8ms
7: E-AC3, 2.0 channels, 192kbit/s, 48khz, dialnorm: -27dB, -27ms

Using graphedt (renamed recode.exe), the Nero splitter only showed one video stream and 3 audio tracks. I was able to play the audio via the Nero Audio Decoder 2.

btw, EVOdemux confirms the two video streams and 4 audio streams, so I'm not sure why the Nero stuff is missing 1 video and the remaining audio stream.

Please let me know what else I can do to help with this one.

May I request a wish list item please. In addition to the video codec, i.e., "VC-1" above as an example, is there a way to provide a few more pieces of info? Something like "VC-1, 1920x1080, 23.976, pulldown" and pulldown would only be there if the video has the pulldown flag. If it's doable.

[EDIT] Well, 5 seconds after I posted this, I tried something else. I renamed eac3tov2.exe to recode.exe and reran the line above and what do you know?

Nero Audio Decoder (Nero 7 or older) works fine
Sonic Audio Decoder (4.3.0.169) works fine
Haali Media Splitter (2007-11-18) is installed
Surcode DTS Encoder (1.0.23.0) is installed
MkvToolnix (v2.1.0) is installed
EVO/VOB, 2 video tracks, 4 audio tracks, 1:29:31
1: Joined EVO/VOB file
2: VC-1
3: VC-1
4: E-AC3, 5.1 channels, 384kbit/s, 48khz, dialnorm: -27dB, -8ms
5: TrueHD, 5.1 channels, 48khz, dialnorm: -24dB, -1ms
6: E-AC3, 2.0 channels, 192kbit/s, 48khz, dialnorm: -27dB, -8ms
7: E-AC3, 2.0 channels, 192kbit/s, 48khz, dialnorm: -27dB, -27ms

[shanghai2004]

Originally Posted by madshi

Yes, that'd be helpful! But only if the problem can be reproduced with the sample, too. Thanks!

http://www.sendspace.com/file/i867cw

Its about 96MB. Plays fine in PowerDVD.
Extracting the LPCM audio track results in corrupted audio file.

Hope you have time to take a look. Thanks ahead!

[yonta]

Samples of 24bit Blu-ray LPCM which crash eac3to v2.10.

All files were demuxed with xport and they seem to be OK when converted to wav with sox/wavewizard.

24bit.2.0ch.Blu-ray.LPCM.sample1
http://www.sendspace.com/file/tnbe4v

24bit.2.0ch.Blu-ray.LPCM.sample2
http://www.sendspace.com/file/ett8j3

24bit.2.0ch.Blu-ray.LPCM.sample3
http://www.sendspace.com/file/5ncjkr

24bit.5.1ch.Blu-ray.LPCM
http://www.sendspace.com/file/64lsvl

Thank you madshi for your great effort!

[madshi]

Originally Posted by nautilus7

Ok, i understand.

Is there anything (except delay ) that would you like me to test?

Nothing specific. Just let eac3to convert some movies for you and check whether everything works alright. That'd be helpful.

[madshi]

Originally Posted by Chumbo

Well, 5 seconds after I posted this, I tried something else. I renamed eac3tov2.exe to recode.exe

Ehm, normally Nero's decoder only works if you rename the exe to "recode.exe". However, eac3to has implemented a hack around this by also allowing "eac3to.exe". Renaming "eac3to.exe" to "eac3tov2.exe" disables this hack. Please don't rename "eac3to.exe".

[madshi]

Originally Posted by Thunderbolt8

can you tell in how far codes or routines from other programs (for example evodemux or h264tsto or haali filters) were used regarding .evo joining & de- and remuxing? would be useful in case we get errors, so we might know which program(s) would produce a similar error in that case and dont have to try each other tool then.

Evo demuxing and joining is completely my own code. I got the idea from Ron's (drmpeg's) original code, but I've totally rewritten it. Timestamp fixing is based on my own OffsetPTS, but also rewritten. Evo remuxing is done on the fly by feeding the joined Evo directly into the "Haali Media Splitter (AR)" without writing the joined Evo file to harddisk first. Timestamp rewriting is done in an extra step by calling mkvtoolnix.

FWIW, I've yesterday converted three h264 HD DVD movies (Shooter, Transformers and Hunt for Red October) with the new eac3to and all three movies converted perfectly (apart from the one remaining libav TrueHD decoder bug). The final MKV files work great and also seeking works great. So I'm feeling safe to say that with the latest Haali filters remuxing h264 HD DVD movies works very well. So no need to demux the h264 video track to a raw file and mux it to MKV by dropping it into mkvtoolnix, anymore. eac3to uses the Haali filters instead which seem now up to the task, as far as I can say so far. I still have some problems with rewriting timestamps with Equilibrium, but I think that's not Haali's fault, but a bug in mkvtoolnix.

[madshi]

Originally Posted by yonta

Samples of 24bit Blu-ray LPCM which crash eac3to v2.10.

Thank you for the samples. The crash should be fixed in the next eac3to build. However, how did you demux these samples? They look kind of corrupt to me. Did you use TsRemux or xport?

[madshi]

Originally Posted by Chumbo

May I request a wish list item please. In addition to the video codec, i.e., "VC-1" above as an example, is there a way to provide a few more pieces of info? Something like "VC-1, 1920x1080, 23.976, pulldown" and pulldown would only be there if the video has the pulldown flag. If it's doable.

That's already planned for a future version.

[madshi]

Originally Posted by shanghai2004

http://www.sendspace.com/file/i867cw

Its about 96MB. Plays fine in PowerDVD.

Does PowerDVD play the PCM track or the DTS-HD track?

The PCM track seems to be corrupt. Haali's splitter doesn't even offer to demux it. Sonic's splitter crashes when trying to demux it. Nero's filter demuxes the track, but it's no valid PCM data. eac3to demuxes it, too, but it's also not valid PCM data.

Maybe EvoDemux corrupted the EVO while rebuilding? Please try eac3to on the original EVO files.

[shambles]

is eac3to supposed to apply the correct delay if you only encode the lossless track from the evos to flac (no video remuxing)? it doesn't seem apply any delay at all..

[nautilus7]

Originally Posted by madshi

Did you use TsRemux or xport?

He mentions xport.

[madshi]

Originally Posted by shambles

is eac3to supposed to apply the correct delay if you only encode the lossless track from the evos to flac (no video remuxing)? it doesn't seem apply any delay at all..

The delay should still be applied correctly. How does the eac3to output look like?

[madshi]

Originally Posted by nautilus7

He mentions xport.

Oooops. Will recheck the samples.

[shambles]

E:\Program Files\eac3to>eac3to 101.evo f:\1.flac
EVO/VOB, 1 video track, 4 audio tracks, 1:11:52
1: VC-1
2: E-AC3, 5.1 channels, 448kbit/s, 48khz, dialnorm: -27dB, 2002ms
3: TrueHD, 5.1 channels, 48khz, dialnorm: -24dB, 2002ms
4: E-AC3, 5.1 channels, 448kbit/s, 48khz, dialnorm: -27dB, 2002ms
5: E-AC3, 5.1 channels, 448kbit/s, 48khz, dialnorm: -27dB, 2002ms
Track 3 is used for destination file "1.flac".
Demuxing 2nd audio track...
Removing dialog normalization...
Encoding FLAC...
Creating/writing file "f:\1.24bit.flac"...
This audio track contains only 16 bit of information.
The zero bytes were successfully removed.
Done.

also, the new eac3to version seems to report dialnorm -24db for all truehd tracks (both in evo and demuxed) while the old ones reported -27db (even for the same tracks)

[madshi]

Originally Posted by shambles

3: TrueHD, 5.1 channels, 48khz, dialnorm: -24dB, 2002ms

Hmmmm... The delay really doesn't seem to be applied. So is the FLAC track out of sync by 2 seconds?

It would be very helpful, if you could upload a little sample of the EVO file. Maybe 50MB, if possible?

Originally Posted by shambles

also, the new eac3to version seems to report dialnorm -24db for all truehd tracks (both in evo and demuxed) while the old ones reported -27db (even for the same tracks)

Ah - thanks for reporting! That's a bug, will be fixed in next build.