eac3to - audio conversion tool

[madshi]

Originally Posted by Snowknight26

Madshi, here is a sample from The Phantom of the Opera:
http://www.stfcc.org/misc/PEVOB_1.EVO

This EVO file seems to be majorly fucked up. The EVO structure is corrupt, so seems to be the audio data. Haven't even looked at the video data. Please rerip the movie and try again.

[madshi]

Originally Posted by nautilus7

That's possible...
I tried the -check16bit switch, but it told me that the track contains more than 16 bit of information. But how many exactly, none knows...

How did the studio manage to make a track with variable bit depth?

Every TrueHD track has a bitdepth of 24bit. *EVERY ONE*. Every TrueHD decoder always outputs 24bit. However, here comes the big key: Sometimes only 16bit of those 24bit are filled with real audio data and the rest is just zeroed out. With Pirates 1 most of the track had only 16bit of the 24bit filled with data, while the remaining 8bit were always zeroes. But there were a few (very few) parts of the audio track where suddenly those 8bit had data in them.

With Inside Man, if it's a 20bit track (which I don't know), the upper 20bit are filled with real data while the lower 4bit are always filled with zeroes.

[madshi]

Originally Posted by Chumbo

I have a short test pcm file that converts just fine to both ac3 and dts, see output below. However, when you run it to just get the info on the file, the program crashes, i.e., "eac3to audio.pcm" crashes.

Will be fixed in the next build. Thanks for reporting.

[nautilus7]

Originally Posted by madshi

Every TrueHD track has a bitdepth of 24bit. *EVERY ONE*. Every TrueHD decoder always outputs 24bit. However, here comes the big key: Sometimes only 16bit of those 24bit are filled with real audio data and the rest is just zeroed out. With Pirates 1 most of the track had only 16bit of the 24bit filled with data, while the remaining 8bit were always zeroes. But there were a few (very few) parts of the audio track where suddenly those 8bit had data in them.

With Inside Man, if it's a 20bit track (which I don't know), the upper 20bit are filled with real data while the lower 4bit are always filled with zeroes.

I decoded the truehd track to wavs. Then i dithered down to 22 and 20 bits. All of them were different, at least at the parts i looked in. Maybe i was unlucky and checked the wrong parts. I 'll check again if i get time.

[shambles]

you'd need to truncate, not dither, the bit depth to compare like that.. but afaik there are no apps that can truncate/convert without dithering to other than the usual 8/16/24/32 bits

[Sephiroth0000]

Madshi need help with EAC3TO please!

I have noticed that when I EVODEMUX my HD DVD it tells me with the audio how many seconds its going to be in or out by. With one certain movie the time is +83ms and when I attempt to do it via the GUI of EAC3TO (yes I have all the nesscary files) it saids its not reconised and it fails.....help!?

[nautilus7]

Originally Posted by shambles

you'd need to truncate, not dither, the bit depth to compare like that.. but afaik there are no apps that can truncate/convert without dithering to other than the usual 8/16/24/32 bits

Sorry, wasn't aware of that.

[madshi]

Originally Posted by shambles

you'd need to truncate, not dither, the bit depth to compare like that.. but afaik there are no apps that can truncate/convert without dithering to other than the usual 8/16/24/32 bits

That's correct. @nautilus7, "dithering" involves adding a certain amount of random noise to an audio track. So obviously after dithering you can't compare audio tracks, anymore. You cannot even compare the audio tracks, if you do the same operation twice. Because of the random noise the result is different everytime.

[madshi]

Originally Posted by Sephiroth0000

Madshi need help with EAC3TO please!

I have noticed that when I EVODEMUX my HD DVD it tells me with the audio how many seconds its going to be in or out by. With one certain movie the time is +83ms and when I attempt to do it via the GUI of EAC3TO (yes I have all the nesscary files) it saids its not reconised and it fails.....help!?

"its" not recognized? What do you mean with "its"? The source EVO file(s)? Or the audio file you demuxed with EvoDemux? Please post the eac3to output here and give me a few more details about what exactly you did.

[shanghai2004]

Just downloaded V2.10 of the program. Impressive!
Now I can directly feed it my EVO files.

But...

Still cannot extract LPCM audio from the Eagles HD-DVD.

E:\codec_mpa>eac3to c:\hddvd\hvdvd_ts\concert1.evo test.flac
EVO/VOB, 1 video track, 2 audio tracks, 1:26:36
1: VC-1
2: RAW/PCM, 2.0 channels, 16 bits, 48khz
3: DTS Master Audio, 5.1 channels, 24 bits, 48khz, dialnorm: -4dB
Track 2 is used for destination file "test.flac".
Demuxing 1st audio track...
Reading RAW/PCM...
Swapping endian...
Encoding FLAC...
Creating/writing file "test.flac"...
Done.

Seems good, but playing back test.flac file sounds like sound is played too slow (low pitch) and with static (digital noise).

EVO plays correctly in PowerDVD...

Helpfull if I supply a sample of the EVO file?

[madshi]

Originally Posted by shanghai2004

Helpfull if I supply a sample of the EVO file?

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

[Sephiroth0000]

Originally Posted by madshi

"its" not recognized? What do you mean with "its"? The source EVO file(s)? Or the audio file you demuxed with EvoDemux? Please post the eac3to output here and give me a few more details about what exactly you did.

DEMUXING HD DVD and using extensions ddp. for the audio. Taking DDP file and putting into EAC3TO GUI and putting output as wav. extension and then attempting to set audio delay to +83ms. Click on CONVERT and then it fails with the line *83 not reconised (cannot remember exactly as not at computer)

[madshi]

Originally Posted by Sephiroth0000

DEMUXING HD DVD and using extensions ddp. for the audio. Taking DDP file and putting into EAC3TO GUI and putting output as wav. extension and then attempting to set audio delay to +83ms. Click on CONVERT and then it fails with the line *83 not reconised (cannot remember exactly as not at computer)

I need the exact eac3to output text. Can't help without that.

[Chumbo]

I wanted to report another issue I just ran into. I'm not sure if what I attempted is correct or not, but basically I wanted to demux the video into an mkv and the specific audio to dts from TrueHD. The process successfully created the mkv and created the WAV files, but never created the resulting DTS. Below is the full output from the process. Note that there was no crash, just what's indicated below. Unfortunately, the created WAV files were deleted. It would be nice to keep those if the process fails.

eac3tov2 FEATURE_1.EVO+FEATURE_2.EVO 2: e:\media\video\movie.mkv 4: movie.dts

EVO/VOB, 1 video track, 4 audio tracks, 2:15:09
1: Joined EVO/VOB file
2: h264/AVC
3: E-AC3, 5.1 channels, 768kbit/s, 48khz, dialnorm: -27dB, -84ms
4: TrueHD, 5.1 channels, 48khz, dialnorm: -24dB, -84ms
5: E-AC3, 5.1 channels, 448kbit/s, 48khz, dialnorm: -27dB, -84ms
6: E-AC3, 5.1 channels, 448kbit/s, 48khz, dialnorm: -27dB, -84ms
Demuxing 2nd audio track...
Removing dialog normalization...
Writing WAVs...
Creating/writing file "movie.L.wav"...
Creating/writing file "movie.R.wav"...
Creating/writing file "movie.LFE.wav"...
Creating/writing file "movie.SL.wav"...
Creating/writing file "movie.C.wav"...
Creating/writing file "movie.SR.wav"...
Muxing video to Matroska...
-------------------------------------------------------------------------------
[mlp @ 68A4D2E0]End of stream indicated
[mlp @ 68A4D2E0]Substream 1 parity check failed
[mlp @ 68A4D2E0]Substream 1 checksum failed
[mlp @ 68A4D2E0]Substream 1 length mismatch.
The libav decoder reported an error while decoding.
Waiting for DirectShow decoder thread to finish. Please wait...

[EDIT] I know this was reported earlier and I'll use the -nero for now. Sorry for the repost. This was with v2.10 btw.

[XolocoTuxmaster]

Thanks for the new update, it's good to know that ffmpeg/libav is really reliable.

What about a linux version now? XD

At least, can you tell me where I can found the patch and how to apply/use it in the linux version of ffmpeg (or libav?)?

[scarbrtj]

Originally Posted by madshi

My first guess would be that the muxer doesn't like the DTS-HD core. There's a small difference to "normal" DTS files. Normal DTS files usually have 2013 bytes per DTS frame while the DTS core from a DTS-HD track only has 2012 bytes per DTS frame. I don't know it behaves this way. And it gets even stranger: If you want to keep audio sync, the DTS parser needs to behave as if the frames were 2013 bytes long!! Extremely strange and kind of annoying. Personally, I'm not muxing the DTS files into the MKV file. Instead I'm keeping them external. The MPC HC can play them as external files. Earlier in this thread you'll find a modified source filter which plays these DTS core tracks with correct audio sync.

As Mr. Spock would say: "Fascinating."

Unfortunately, I must mux, for I must use Windows Media Center

When you talk of this filter... do you mean it will play the DTS core file properly separately, or play it properly muxed? And is the filter dtsac3source.ax?

[scarbrtj]

Originally Posted by madshi

My first guess would be that the muxer doesn't like the DTS-HD core. There's a small difference to "normal" DTS files. Normal DTS files usually have 2013 bytes per DTS frame while the DTS core from a DTS-HD track only has 2012 bytes per DTS frame. I don't know it behaves this way. And it gets even stranger: If you want to keep audio sync, the DTS parser needs to behave as if the frames were 2013 bytes long!! Extremely strange and kind of annoying. Personally, I'm not muxing the DTS files into the MKV file. Instead I'm keeping them external. The MPC HC can play them as external files. Earlier in this thread you'll find a modified source filter which plays these DTS core tracks with correct audio sync.

Oooh! Posting twice in a row. So embarassing. BUT... might one hypothesize (to avoid the "2012/2013" bug) that if one went:

DTS-HD --> mono wavs --> SurCode DTS encode --> mux AVC and DTS

versus:

DTS-HD --> DTS core --> mux AVC and DTS

that the former would be more ideal than the latter? To get a "normal" DTS track?

[madshi]

Originally Posted by XolocoTuxmaster

Thanks for the new update, it's good to know that ffmpeg/libav is really reliable.

Well, not fully yet, but we're working on it. It seems there's one last annoying bug, but I hope it'll be sorted out soon.

Originally Posted by XolocoTuxmaster

What about a linux version now? XD

No chance in hell, sorry.

[madshi]

Originally Posted by scarbrtj

When you talk of this filter... do you mean it will play the DTS core file properly separately, or play it properly muxed? And is the filter dtsac3source.ax?

The filter only works for separate files and yes, I mean the modified dtsac3source.ax.

Originally Posted by scarbrtj

might one hypothesize (to avoid the "2012/2013" bug) that if one went:

DTS-HD --> mono wavs --> SurCode DTS encode --> mux AVC and DTS

versus:

DTS-HD --> DTS core --> mux AVC and DTS

that the former would be more ideal than the latter? To get a "normal" DTS track?

Yes, transcoding the DTS-HD track to something else should help. You can use the Surcode DTS encoder or you could also use FLAC (maybe dithered down to 18 bits?).

[nautilus7]

madshi, i'm sorry. I don't have any hd dvd with a lossless track that needs a delay.
Actually, i remember adding a delay to letters from iwo jima hd dvd when i was making a flac some weeks ago, so i tried that movie.
But, unfortunately, that was due the eac3to bug that caused truehd tracks to be out of sync (no delay needed), so nothing helpful to report.

But i want to ask something...

I typed:

eac3to feature1.evo+feature2.evo 4: output1.flac 4: output2.ac3

where track 4 is a truehd audio track. 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?