eac3to - audio conversion tool - Page 425

lchiu7 · 8th March 2009, 03:42

Quote:

Originally Posted by peterjcat

You got it. Most DD+ tracks on HD DVD are 1.536mbps, to unpack that to 24-bit 5.1-channel LPCM take 6.9mbps, so it's 4.5x the size to maintain the same quality. Even if you've got plenty of disc space, it's the principle of the thing!

Well the AQ drop to 640Kbs DD shouldn't be that noticeable. Now if we can re-encode as TrueHD that would solve the problems since I think I see Madshi saying there isn't an affordable DD+ encoder around

tebasuna51 · 8th March 2009, 04:10

Quote:

Originally Posted by Snowknight26

eac3to can't properly decode this AC3 file to WAV.

That file have the first 136 frames 2.0 and the rest 5.1

Use delaycut to split the file.

Snowknight26 · 8th March 2009, 04:45

Hah, I would have never guessed.

Thanks the the advice, worked perfectly.

ACrowley · 8th March 2009, 09:55

Quote:

Originally Posted by tebasuna51

That file have the first 136 frames 2.0 and the rest 5.1

Use delaycut to split the file.

yep..it works.

I had such AC3 File yesterday. Its a AC3 from a DVB-C HD Capture. The First Frame/032ms is 2.0.
eac3to/tsmuxer etc cant detect the AC3.

Cutting out 32ms works perfect

idbirch2 · 8th March 2009, 11:42

Quote:

Originally Posted by madshi

No. It does not.

It doesn't change one bit in the DD+ stream. It just slightly changes the way the DD+ stream is stored in the m2ts container.

OK, I'm sorry, I just assumed that seeing as I processed an HD-DVD DD+ track using QuickFix.cmd (which I'm sure is part of ToNMT) and it then played on my PS3 as DD+ that it must be doing some sort of conversion. All I know is, you can play DD+ from a HD-DVD on a PS3 in the AVCHD format.

madshi · 8th March 2009, 14:52

Quote:

Originally Posted by Kurtnoise

@madshi: could you add Dirac detection from TS files please ?

Is this actually used in real life anywhere?

Quote:

Originally Posted by GZZ

I think there is a small bug in EAC3to way of parsing the mpls files and the m2ts files.

I did a test on Der Untergang (Nordic version)

The missing audio tracks and track languages are caused by the very strange way this Blu-Ray is authored. I could probably find a way to fix it, but honestly, I don't care much, because it's the only movie which shows this problem and fixing the problem would require a major rewrite of the playlist/clipinfo reading logic.

Quote:

Originally Posted by GZZ

There is also a problem with extracting video and audio from this using the mpls file. Because the audio goes out of sync because of the small 00010.m2ts black screen file.

This will be fixed in the next build.

madshi · 8th March 2009, 14:58

Quote:

Originally Posted by jmonier

Did you try checking it with something other than eac3to?

No. And I see no reason to do that.

Quote:

Originally Posted by jmonier

In my case, AC3Filter reported it as 2.1 and played it that way as well.

So maybe AC3Filter has a bug? Or maybe the WAV source filter doesn't forward the channel mask?

Quote:

Originally Posted by jmonier

And when I fed the WAV directly to my receiver via SPDIF the center channel was not there.

No surprise there. SPDIF only supports 2 channels PCM/WAV transport.

Quote:

Originally Posted by DrNein

But the problem appears to occur in the conversion before tsMuxeR is involved as the output plays wrong on its own.

The eac3to output plays wrong on its own? How did you test that?

Quote:

Originally Posted by DrNein

In PDVD the track is identified as MLP rather than TrueHD as by eac3to.

Which demonstrates that PDVD is not a tool you should trust.

------

Guys, I'm very much willing to fix bugs in eac3to. But I'm not going to spend hours and hours on trying to figure out which of the various tools you're using is responsible for a specific problem you have. If you want me to look into a problem then you should better provide some convincing proof that the problem is actually caused by eac3to.

madshi · 8th March 2009, 15:46

eac3to v3.13 released

http://madshi.net/eac3to.zip

Code:

* fields and frames are counted and displayed separately now
* added DIRAC bitstream parser
* added support for "-24.975" and "-changeto24.975"
* Blu-Ray subtitle demuxing: PTS value is now written to both PTS + DTS
* joining MKV files is now declined with a proper error message
* last chapter is now removed, if it's less than 10 seconds from end of movie
* fixed: "-normalize" didn't work with stdout, anymore
* fixed: audio delay was incorrect when 1st m2ts part contained no audio data
* fixed: very small WAV files were not detected correctly
* fixed: "eac3to source.eac3 dest.dts -core" crashed

madshi · 8th March 2009, 15:52

Quote:

Originally Posted by anode

i have a short question to subtitle demuxing:

as far i have seen, in the output of eac3to's demuxed sup-files from blu-ray they contain the necessary PTS info but no DTS (decoder time stamp) information. I had some subtitles which do not play on my Samsung BD-P2500 (longer, more complex ones) (remuxed with tsMuxeR to BR-AVCHD). Why are the DTS fields always zero?
When demuxing the same track with tsMuxeR, DTS info is present.

If i use the tsMuxeR-demuxed source, then the subtitles are displayed correctly.
Perhaps you can give me some info about this DTS information?

Can you please retry with v2.13? I've done a quick and dirty fix by simply copying the PTS value to the DTS field. I hope that this solves the problem. If not, I'll have to do a proper (and time consuming) fix.

TinTime · 8th March 2009, 16:07

Quote:

Originally Posted by madshi

eac3to v2.13 released

Thanks very much!

Just one thing - it's v3.13 isn't it?

DrNein · 8th March 2009, 16:12

Quote:

Originally Posted by madshi

The eac3to output plays wrong on its own? How did you test that?

The extracted THD is not identified by MediaInfo nor can it be played in MPC-HC with decoders such as CyberLink. The converted THD+AC3 can be be but the duration and playback is wrong (or presumably the playback is "accurate" but the conversion is screwy). The extracted THD can be played with a plugin for Winamp based upon mlpdec module from FFmpeg and as said converts okay to DTS.

Quote:

Originally Posted by madshi

Which demonstrates that PDVD is not a tool you should trust.

I thought it might demonstate that the THD track was odd and thus the dodgy conversion to THD+AC3.

But anyway it is not really worth spending anymore time on -especially as the conversion to DTS is fine. Again, I do appreciate your time and excellent tool.

Oops, typo in the new version release above... should be v3.13 -madshi needs a well deserved break

madshi · 8th March 2009, 16:55

Quote:

Originally Posted by DrNein

The extracted THD is not identified by MediaInfo nor can it be played in MPC-HC with decoders such as CyberLink.

Isn't that true for every THD+AC3 track?

Quote:

Originally Posted by DrNein

The converted THD+AC3 can be be but the duration and playback is wrong

Isn't that true for every THD+AC3 track? TrueHD is a variable bitrate codec, so duration is almost always shown wrong. The source filter only does a best guess on the duration.

Quote:

Originally Posted by DrNein

I thought it might demonstate that the THD track was odd and thus the dodgy conversion to THD+AC3.

No, not at all. Most probably PDVD always reports MLP for every THD+AC3 track.

laserfan · 8th March 2009, 16:59

Quote:

Originally Posted by madshi

eac3to v3.13 released

http://madshi.net/eac3to.zip

Code:

* fields and frames are counted and displayed separately now

Thanks for the update. I was kinda hoping it would ID "hybrid" clips which have a mix of I/L and progressive frames, but instead I will continue to rely on neuron2's tools for that. Here's a nasty one, it's actually 29.97:

Code:

eac3to v3.13
command line: eac3to E:\MAMMA_MIA!\BDMV\STREAM\00289.m2ts 1: test.vc1
------------------------------------------------------------------------------
M2TS, 1 video track, 1 audio track, 3 subtitle tracks, 0:02:40, 60i /1.001
1: VC-1, 1080i60 /1.001 (16:9)
2: DTS Master Audio, English, 5.1 channels, 24 bits, 48khz, dialnorm: -4dB
   (core: DTS, 5.1 channels, 24 bits, 1509kbps, 48khz, dialnorm: -4dB)
3: Subtitle (PGS), English
4: Subtitle (PGS), Spanish
5: Subtitle (PGS), French
[v01] Extracting video track number 1...
[v01] Creating file "test.vc1"...
Video track 1 contains 4800 frames.
eac3to processing took 11 seconds.
Done.

DrNein · 8th March 2009, 18:13

Quote:

Originally Posted by madshi

Isn't that true for every THD+AC3 track?

No, this specific THD (at least) cannot be identified by MediaInfo or played by MPC-HC but after conversion to THD+AC3 it can. As far as I recall, other unaltered THD+AC3 (such as from BD) can likewise be identified and played but I would have to check more to confirm. Would have to check PDVD identification as well.

The duration of the THD+AC3 (converted) is crazy long and the counter plays fast. Plus as said the audio drops out or in other words is a bit warbly.

As for the BD subtitle problem, would that affect their display in MPC-HC? Because I notice that whether they are shown or not is unreliable and sometimes takes toggling off/on in the menu under subtitles and under splitter, but I assumed that was the fault of the internal splitter. Also, did the same problem exist with extraction from HD DVD or is that not applicable? Is there any way to fix already extracted SUP files (copy PTS value to DTS -if that turns out to be the fix)?

peterjcat · 8th March 2009, 22:28

Quote:

Originally Posted by lchiu7

Well the AQ drop to 640Kbs DD shouldn't be that noticeable. Now if we can re-encode as TrueHD that would solve the problems since I think I see Madshi saying there isn't an affordable DD+ encoder around

No, it shouldn't be THAT noticeable, but again it's the principle of the thing

And sure, recoding to TrueHD would be great, but no more affordable: you need the same program from Dolby. Surcode has an MLP encoder but I don't know if there's any way to convert the output to useful TrueHD.

madshi · 8th March 2009, 23:20

Quote:

Originally Posted by laserfan

Thanks for the update. I was kinda hoping it would ID "hybrid" clips which have a mix of I/L and progressive frames

What is "I/L"?

eac3to doesn't handle interlaced VC-1 streams well in the moment. That's still on my to do list. Frame/field counting should work fine for h264 and MPEG2, though.

Quote:

Originally Posted by DrNein

No, this specific THD (at least) cannot be identified by MediaInfo or played by MPC-HC but after conversion to THD+AC3 it can.

So probably *NO* THD (without AC3) can be detected by MediaInfo or MPC-HC. Have you checked that? Probably no. But you should before reporting a problem.

Quote:

Originally Posted by DrNein

The duration of the THD+AC3 (converted) is crazy long and the counter plays fast.

Incorrect duration probably happens for most/all THD+AC3 tracks. Have you checked that? Probably no. But you should before reporting a problem.

These are most probably not eac3to problems. So let's please finally stop discussing this. Unless you find some proof for an eac3to bug.

Quote:

Originally Posted by DrNein

As for the BD subtitle problem, would that affect their display in MPC-HC?

Why don't you try for yourself?

laserfan · 9th March 2009, 00:22

Quote:

Originally Posted by madshi

What is "I/L"? eac3to doesn't handle interlaced VC-1 streams well in the moment. That's still on my to do list.

Sorry. Interlaced, of course! And my post was indeed about a VC-1 clip. If you can detect/identify the messes that are these hybrid VC-1 clips, you will join neuron2 as the only who's figured these out AFAICT.

TM2-Megatron · 9th March 2009, 01:37

This seems like a very useful tool, however I'm not sure about a few of these additional programs or plug-ins that are required for some of its most useful features.

The Nero HD-DVD/Blu-Ray plug-in isn't even available anymore, so I'm not sure how people who didn't buy it during the relatively short window it was for sale would get it. As for the Arcsoft plug-in, I'm not sure which of their programs this one comes with, however I haven't heard great things about their software so I wouldn't really want to pay for it anyway. And Surcode's DTS plug-in is ridiculously expensive.

Are there any alternatives to some of these impractical plug-ins?

TinTime · 9th March 2009, 05:27

Madshi

Sorry to do this to you but there seems to be a problem introduced in 3.13. The output audio is corrupted when the input file is a wav.

I ran...
"D:\Vtemp\programs\eac3to\eac3to.exe" mh2.wav mhtest2.flac
...with this test file. It's processed correctly in 3.12.

NB if you play back the corrupt output file turn your speakers down. The noise is pretty loud

Thank you.

Snowknight26 · 9th March 2009, 08:59

Quote:

C:\unzipped\eac3to>eac3to C:\temp\test.wav test.flac
WAV, 1.0 channels, 0:00:30, 64 bits <float>, 6144kbps, 96khz
Reading WAV...
eac3to crashed...

C:\unzipped\eac3to>eac3to C:\temp\test2.wav test2.flac
WAV, 1.0 channels, 0:00:30, 32 bits <float>, 3072kbps, 96khz
Reading WAV...
eac3to crashed...

C:\unzipped\eac3to>eac3to.exe C:\temp\test3.wav C:\temp\test3.flac
WAV, 1.0 channels, 0:00:30, 32 bits, 3072kbps, 96khz
Reading WAV...
Reducing depth from 32 to 24 bits...
Encoding FLAC with libFlac...
Creating file "test3.flac"...
Original audio track: max 32 bits, average 28 bits, most common 27 bits.
The processed audio track has a constant bit depth of 24 bits.
eac3to processing took 1 second.
Done.

C:\unzipped\eac3to>eac3to.exe C:\temp\test3.wav C:\temp\test3.down24.try2.flac -down24
WAV, 1.0 channels, 0:00:30, 32 bits, 3072kbps, 96khz
Reading WAV...
Reducing depth from 32 to 24 bits...
Encoding FLAC with libFlac...
Creating file "test3.try2.flac"...
Original audio track: max 32 bits, average 28 bits, most common 27 bits.
The processed audio track has a constant bit depth of 24 bits.
eac3to processing took 1 second.
Done.

C:\unzipped\eac3to>eac3to.exe C:\temp\test3.wav C:\temp\test3.down24.flac -down24
WAV, 1.0 channels, 0:00:30, 32 bits, 3072kbps, 96khz
Reading WAV...
Reducing depth from 32 to 24 bits...
Encoding FLAC with libFlac...
Creating file "C:\temp\test3.down24.flac"...
Original audio track: max 32 bits, average 28 bits, most common 27 bits.
The processed audio track has a constant bit depth of 24 bits.
eac3to processing took 1 second.
Done.

C:\unzipped\eac3to>eac3to.exe C:\temp\test3.wav C:\temp\test3.full.flac -full
WAV, 1.0 channels, 0:00:30, 32 bits, 3072kbps, 96khz
Reading WAV...
Reducing depth from 32 to 24 bits...
Encoding FLAC with libFlac...
Creating file "C:\temp\test3.full.flac"...
Original audio track: max 32 bits, average 28 bits, most common 27 bits.
The processed audio track has a constant bit depth of 24 bits.
eac3to processing took 1 second.
Done.

Couple of things to notice:
32-bit and 64-bit floating point WAV files made with Audacity crash eac3to
32-bit doesn't.
All the FLAC files have differen't sizes and aren't identical to test3.wav. Notice the repeated sound in the beginning.
I've included a reference FLAC file named audacity.flac.

http://www.stfcc.org/misc/wavs.zip (51,803,165 bytes) - should be finished uploading within 30 minutes of posting this.

8th March 2009, 04:45	#8483 \| Link
Snowknight26 Registered User Join Date: Aug 2007 Posts: 1,430	Hah, I would have never guessed. Thanks the the advice, worked perfectly.

9th March 2009, 01:37	#8498 \| Link
TM2-Megatron Registered User Join Date: Mar 2007 Posts: 5	This seems like a very useful tool, however I'm not sure about a few of these additional programs or plug-ins that are required for some of its most useful features. The Nero HD-DVD/Blu-Ray plug-in isn't even available anymore, so I'm not sure how people who didn't buy it during the relatively short window it was for sale would get it. As for the Arcsoft plug-in, I'm not sure which of their programs this one comes with, however I haven't heard great things about their software so I wouldn't really want to pay for it anyway. And Surcode's DTS plug-in is ridiculously expensive. Are there any alternatives to some of these impractical plug-ins?

9th March 2009, 05:27	#8499 \| Link
TinTime Registered User Join Date: Jan 2009 Location: UK Posts: 403	Madshi Sorry to do this to you but there seems to be a problem introduced in 3.13. The output audio is corrupted when the input file is a wav. I ran... "D:\Vtemp\programs\eac3to\eac3to.exe" mh2.wav mhtest2.flac ...with this test file. It's processed correctly in 3.12. NB if you play back the corrupt output file turn your speakers down. The noise is pretty loud Thank you.