Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > General > Audio encoding

Closed Thread
 
Thread Tools Search this Thread Display Modes
Old 16th February 2009, 01:05   #8281  |  Link
DrNein
Registered User
 
Join Date: Sep 2002
Posts: 145
Quote:
Originally Posted by madshi View Post
A while ago I tested the Nero decoder by feeding it an E-AC3 test frame. Then I thought about adding support for the Nero 9 AC3 decoder, which however can't decode E-AC3, anymore, but only AC3. So I changed the Nero decoder test to decode an AC3 test frame. But in the meanwhile I've decided that adding Nero 9 decoder support doesn't make much sense without E-AC3 support, so I recently switched back to testing E-AC3 decoding instead of AC3 decoding. I don't remember exactly in which version I changed that (it's not in the changelog, I think), but it could have been v3.06. So my best guess would be that AC3 decoding works on your PC, but E-AC3 decoding does not produce the expected results. Don't ask me why, though. As I said, on my PC it works just fine.
Okay, I will not ask you why... but can I ask you what dependencies NeAudio2.ax has for E-AC3 decoding other than NeEacDec.dll?

Even with the complete Audio Plugins, DSFilter, and Lib folders restored, it is still not working with eac3to 3.06-3.11. Perhaps another DSFilter ax needs registering or there is some other registry entry missing?

I see what you mean about that period where it behaved differently: 2.87 is the oldest version I still have which works on my system (so, presumably at least 2.87-3.05) while 2.80 does not.

Last edited by DrNein; 16th February 2009 at 01:19.
DrNein is offline  
Old 16th February 2009, 02:15   #8282  |  Link
73ChargerFan
Registered User
 
73ChargerFan's Avatar
 
Join Date: Dec 2006
Posts: 523
Quote:
Originally Posted by Snowknight26 View Post
All files. I'm mostly running eac3to over a gigabit network, but it happens even when running it directly to/from an incredibly fast RAID. As soon as I type any command, eac3to sits there for several seconds before even showing the progress bar.
Just a thought, but check to see if your antivirus is scanning the m2ts or other audio/video files.

I had to exclude file extensions, my media drives and the working directories I use for video conversions/muxing.
73ChargerFan is offline  
Old 16th February 2009, 02:22   #8283  |  Link
odin24
Registered User
 
odin24's Avatar
 
Join Date: Mar 2008
Location: The Great North (the better half of North America)
Posts: 301
Does eac3to still detect forced subs on BDs? If so it didn't on Iron Man (North American).

Thanks,
odin24 is offline  
Old 16th February 2009, 02:39   #8284  |  Link
Snowknight26
Registered User
 
Join Date: Aug 2007
Posts: 1,430
Quote:
Originally Posted by 73ChargerFan View Post
I had to exclude file extensions, my media drives and the working directories I use for video conversions/muxing.
I have nothing aside from vital OS processes/services running on the machine that exhibited the behavior. But oh well, it seems to have gone away for the most part with 3.11.
Snowknight26 is offline  
Old 16th February 2009, 07:46   #8285  |  Link
kypec
User of free A/V tools
 
kypec's Avatar
 
Join Date: Jul 2006
Location: SK
Posts: 826
Hi madshi, I tried that -simple switch as you suggested. It works, however, I found out that the WAV files produced with that option are not bit-to-bit identical (data chunk wise). How is that possible? Here are my sample files, which I used for the test. I'd really appreciate if somebody else could test it also and post his report on different/identical AC3->WAV results.
Command lines invoked ->
Modern WAV output (WAVE_FORMAT_EXTENSIBLE)
Code:
T:\eac3to\eac3to.exe "test.VOB" "R:\test.wav" -down16 -libav
Compatible WAV output (old style PCM header)
Code:
T:\eac3to\eac3to.exe "test.VOB" "R:\simple.wav" -down16 -libav -simple
In the following pictures are highlighted just a few discrepancies between those two output files.

In general, they differ a LOT, the samples do not match although they should because they were produced from the same source, by using the same tool and same decoder library. How come they're not identical is beyond my comprehension.
When I imagine all the guys who struggle to get LOSSLESS audio from their TrueHD/DTS-MA and so on while there seem to be weird issues even with ordinary stereo AC3...
kypec is offline  
Old 16th February 2009, 08:39   #8286  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by DrNein View Post
Okay, I will not ask you why... but can I ask you what dependencies NeAudio2.ax has for E-AC3 decoding other than NeEacDec.dll?

Even with the complete Audio Plugins, DSFilter, and Lib folders restored, it is still not working with eac3to 3.06-3.11. Perhaps another DSFilter ax needs registering or there is some other registry entry missing?
I don't really know. I installed full Nero 7, registered proper (actually bought by me) license codes for Nero 7 and for the Blu-Ray/HD DVD plugin, and Nero 7 decoding works just fine for me, without having to manually register any filters.

In the past people only had problems if they tried getting around installing full Nero 7 and/or if they tried to use stolen license keys. I can't - and don't want to - help doing either of that. Now of course I don't know if you have installed full Nero 7 and if you have proper license keys. If you haven't, please don't ask me for help. If you have, I'm sorry about the problem you have, but I don't really know how to help, since everything works just fine on my PC.

Quote:
Originally Posted by odin24 View Post
Does eac3to still detect forced subs on BDs? If so it didn't on Iron Man (North American).
I think it should still work. Not all Blu-Rays do have subs marked as "forced". Please also note that eac3to only knows after having processed the whole source file.

Quote:
Originally Posted by kypec View Post
I tried that -simple switch as you suggested. It works, however, I found out that the WAV files produced with that option are not bit-to-bit identical (data chunk wise). How is that possible?
Because the "simple" header is shorter than the "extensible" one. So you simply cannot compare the files like you did. Your comparison program compared byte by byte without accounting for the different header sizes. Here's a test you can do:

Ask eac3to to convert the "-simple" wav file to extensible by doing "eac3to simple.wav full.wav". And ask eac3to to convert the extensible wav file to simple by doing "eac3to extensible.wav simplified.wav -simple". Then compare "simple.wav" with "simplified.wav" and compare "full.wav" with "extensible.wav" and you should notice that both comparisons are 100% bit identical.

Quote:
Originally Posted by kypec View Post
When I imagine all the guys who struggle to get LOSSLESS audio from their TrueHD/DTS-MA and so on while there seem to be weird issues even with ordinary stereo AC3...
There are no weird issues. Your testing method is invalid.
madshi is offline  
Old 16th February 2009, 10:10   #8287  |  Link
gigah72
Registered User
 
Join Date: Feb 2002
Location: OF/Germany
Posts: 111
the languages are missing

3.08
Code:
c:\_sysprg\staxrip\eac3to>"c:\_sysprg\staxrip\eac3to\eac3to.exe" "C:\_data\_vd\_
Full\MATRIX_HDDVD" 1)
EVO, 2 video tracks, 8 audio tracks, 5 subtitle tracks, 2:16:17
"feature"
1: Joined EVO file
2: Chapters, 38 chapters with names
3: VC-1, 1080p24 /1.001 (16:9) with pulldown flags
4: VC-1, 480p30 /1.001 (3:2), 66ms
5: E-AC3, English, 5.1 channels, 640kbps, 48khz, dialnorm: -27dB
6: TrueHD, English, 5.1 channels, 48khz, dialnorm: -27dB
7: E-AC3, German, 5.1 channels, 640kbps, 48khz, dialnorm: -27dB
8: E-AC3 Surround, English, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB
   "Commentary 1"
9: E-AC3 Surround, English, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB
   "Commentary 2"
10: E-AC3 Surround, English, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB
    "Commentary 3"
11: E-AC3 Surround, English, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB
    "Commentary 4"
12: E-AC3, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB, -3ms
13: Subtitle (DVD), English
14: Subtitle (DVD), German
15: Subtitle (DVD), German, "SDH"
16: Subtitle (DVD)
17: Subtitle (DVD)
3.11
Code:
c:\_sysprg\staxrip\eac3to>"c:\_sysprg\staxrip\eac3to\eac3to.exe" "C:\_data\_vd\_
Full\MATRIX_HDDVD" 1)
EVO, 2 video tracks, 8 audio tracks, 5 subtitle tracks, 2:16:17
1: Joined EVO file
2: Chapters, 38 chapters with names
3: VC-1, 1080p24 /1.001 (16:9) with pulldown flags
4: VC-1, 480p30 /1.001 (3:2), 66ms
5: E-AC3, 5.1 channels, 640kbps, 48khz, dialnorm: -27dB
6: TrueHD, 5.1 channels, 48khz, dialnorm: -27dB
7: E-AC3, 5.1 channels, 640kbps, 48khz, dialnorm: -27dB
8: E-AC3 Surround, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB
9: E-AC3 Surround, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB
10: E-AC3 Surround, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB
11: E-AC3 Surround, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB
12: E-AC3, 2.0 channels, 192kbps, 48khz, dialnorm: -27dB, -3ms
13: Subtitle (DVD)
14: Subtitle (DVD)
15: Subtitle (DVD)
16: Subtitle (DVD)
17: Subtitle (DVD)
gigah72 is offline  
Old 16th February 2009, 10:41   #8288  |  Link
T800
Registered User
 
Join Date: Jan 2009
Posts: 13
Quote:
Originally Posted by madshi View Post
Can somebody else please try this? Can you reproduce the problem with the sample T800 uploaded?

@T800, you didn't replace any files in the eac3to folder, did you? E.g. libAften DLL or something like that? Please try deleting the whole eac3to folder and redownloading it, just to be safe... For me the samples you uploaded definitely convert just fine!
Just to let you know, v3.11 works fine for me now.
T800 is offline  
Old 16th February 2009, 11:58   #8289  |  Link
kypec
User of free A/V tools
 
kypec's Avatar
 
Join Date: Jul 2006
Location: SK
Posts: 826
What the hell is going on here?

Quote:
Originally Posted by madshi View Post
Because the "simple" header is shorter than the "extensible" one. So you simply cannot compare the files like you did. Your comparison program compared byte by byte without accounting for the different header sizes.
You apparently didn't look at my comparison results closely. I am aware of the difference in header part of the WAV files. I know that EXTENSIBLE header is longer (by 24 bytes precisely) and I have taken this difference into account for my binary comparison. The problem is that RAW SAMPLE DATA are different which I have clearly marked in the pictures.

Quote:
Originally Posted by madshi View Post
Here's a test you can do:

Ask eac3to to convert the "-simple" wav file to extensible by doing "eac3to simple.wav full.wav". And ask eac3to to convert the extensible wav file to simple by doing "eac3to extensible.wav simplified.wav -simple". Then compare "simple.wav" with "simplified.wav" and compare "full.wav" with "extensible.wav" and you should notice that both comparisons are 100% bit identical.

There are no weird issues. Your testing method is invalid.
I did as you proposed and the findings were even more confusing now. I think there must be definitely something wrong with my system. I wrote a batch file in order to prove whether my suspicions are right:
Code:
T:\eac3to\eac3to.exe test.vob test1.wav -libav -down16
T:\eac3to\eac3to.exe test.vob test2.wav -libav -down16
T:\eac3to\eac3to.exe test.vob test3.wav -libav -down16
T:\eac3to\eac3to.exe test.vob test4.wav -libav -down16
T:\eac3to\eac3to.exe test.vob test5.wav -libav -down16
exit
Here comes the big surprise: NO two files of those 5 runs are 100% identical. You can download whole bunch of 5 outputs with MD5 checksums. Each of these 5 files is UNIQUE!
Can anybody else confirm eac3to produces consistent output with WAV files (using built-in libav only). You can use my test.VOB file for your tests.
My system specs: Athlon 64 X2 4200+, Windows XP SP2 32-bit

Desperately waiting for any possible explanation of this abnormal behaviour...
kypec is offline  
Old 16th February 2009, 12:19   #8290  |  Link
honai
Guest
 
Posts: n/a
It's not a bug, it's a feature.

madshi uses random noise dithering for downsampling (instead of a fixed noise pattern in previous 2.x or 1.x versions).

Actually, you have successfully proved that the random number generator of eac3to works correctly.
 
Old 16th February 2009, 12:35   #8291  |  Link
kypec
User of free A/V tools
 
kypec's Avatar
 
Join Date: Jul 2006
Location: SK
Posts: 826
Quote:
Originally Posted by honai View Post
It's not a bug, it's a feature.

madshi uses random noise dithering for downsampling (instead of a fixed noise pattern in previous 2.x or 1.x versions).

Actually, you have successfully proved that the random number generator of eac3to works correctly.
Is there some switch to force eac3to NOT TO use random noise dithering? Or even better, to disable any dithering at all? I think basic mathematical rounding (64..24-bit -> 16-bit) would do just great for me. The problem is that documentation (on-screen usage help) is rather brief and lacking many undocumented options e.g. -simple. I wonder why madshi is hiding them from us?

I tried omit -down16 option as to avoid downconverting. Resulting WAVs (24-bit) are still not consistent. Isn't AC3 internally only 24-bit also? I thought lossy format means some info is lost during encoding, not generated randomly on each decoding.
kypec is offline  
Old 16th February 2009, 12:52   #8292  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by gigah72 View Post
the languages are missing

Code:
eac3to.exe "C:\_data\_vd\_
Full\MATRIX_HDDVD" 1)
Argh, now languages are working for Blu-Ray and MKV, but not for HD DVD. Will be fixed in the next build (probably next Sunday).

Quote:
Originally Posted by kypec View Post
Is there some switch to force eac3to NOT TO use random noise dithering? Or even better, to disable any dithering at all? I think basic mathematical rounding (64..24-bit -> 16-bit) would do just great for me.
If you like audio artifacts, then yes, it would be great for you.

Quote:
Originally Posted by kypec View Post
I tried omit -down16 option as to avoid downconverting. Resulting WAVs (24-bit) are still not consistent. Isn't AC3 internally only 24-bit also?
If you use libav for AC3 decoding, eac3to gets 32bit floating point audio data from libav. AC3 is not stored in integer samples, but it's stored in the time domain. If you want to get "consistent" (the name doesn't fit) decoding output, you can use the undocumented "-full" switch. That will give you floating point WAV files for libav AC3 decoding. Or you can use Nero or Sonic for AC3 decoding (if you have the decoders installed) which internally downsample to 24bit. Don't know which downsample algorithm they use, maybe they use a fixed noise pattern, in that case the output will be "consistent".

BTW, "consistent" is actually bad when talking about reducing audio bitdepth.
madshi is offline  
Old 16th February 2009, 13:14   #8293  |  Link
buzzqw
HDConvertToX author
 
Join Date: Nov 2003
Location: Cesena,Italy
Posts: 6,552
Hi madshi!

is possible to specify what track to extract ?

the -demux seems to extract all.. i would like to extract only audio, or subs, or specific stream/track

thanks!

BHH
__________________
HDConvertToX: your tool for BD backup
MultiX264: The quick gui for x264
AutoMen: The Mencoder GUI
AutoWebM: supporting WebM/VP8
buzzqw is offline  
Old 16th February 2009, 13:20   #8294  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by buzzqw View Post
is possible to specify what track to extract ?

the -demux seems to extract all.. i would like to extract only audio, or subs, or specific stream/track
Yes, of course. You have full control over what gets demuxed, transcoded, whatever. E.g. you can do:

Code:
eac3to source.m2ts 1: video.h264 2: English.ac3 3: French.ac3 8: English.sup
In order to find out which track has which track number, you need to do a track listing first by doing "eac3to source.m2ts".

This logic applies to all supported source formats.

Last edited by madshi; 16th February 2009 at 13:22.
madshi is offline  
Old 16th February 2009, 13:28   #8295  |  Link
hubblec4
Matroska find' ich toll
 
Join Date: Apr 2008
Posts: 1,370
@madshi

great thanks for the updates.

mkv-support is almost ready. i miss the support for subtitle vob.sub/idx.

can you realize that??

hubble
hubblec4 is offline  
Old 16th February 2009, 13:31   #8296  |  Link
buzzqw
HDConvertToX author
 
Join Date: Nov 2003
Location: Cesena,Italy
Posts: 6,552
yup ! thansk madshi

(it's time to update the command line help... )

BHH
__________________
HDConvertToX: your tool for BD backup
MultiX264: The quick gui for x264
AutoMen: The Mencoder GUI
AutoWebM: supporting WebM/VP8
buzzqw is offline  
Old 16th February 2009, 13:35   #8297  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by hubblec4 View Post
mkv-support is almost ready. i miss the support for subtitle vob.sub/idx.

can you realize that??
It's on my to do list, along with chapters and attachments.
madshi is offline  
Old 16th February 2009, 13:47   #8298  |  Link
kypec
User of free A/V tools
 
kypec's Avatar
 
Join Date: Jul 2006
Location: SK
Posts: 826
Quote:
Originally Posted by madshi View Post
If you use libav for AC3 decoding, eac3to gets 32bit floating point audio data from libav. AC3 is not stored in integer samples, but it's stored in the time domain. If you want to get "consistent" (the name doesn't fit) decoding output, you can use the undocumented "-full" switch. That will give you floating point WAV files for libav AC3 decoding. Or you can use Nero or Sonic for AC3 decoding (if you have the decoders installed) which internally downsample to 24bit. Don't know which downsample algorithm they use, maybe they use a fixed noise pattern, in that case the output will be "consistent".

BTW, "consistent" is actually bad when talking about reducing audio bitdepth.
I see, after a bit of study in A/52 documentation about noise dither during decoding. Could you please elaborate why "consistency" (meaning to obtain identical results when decoding through same tools from the same source) is bad? I was so sure until today, that SAME BINARY SOURCE DATA->SAME PROCESSING ALGORITHM->SAME BINARY OUTPUT in the world of computing...well...I'll just have to live with that being not so true always
kypec is offline  
Old 16th February 2009, 13:57   #8299  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by kypec View Post
I see, after a bit of study in A/52 documentation about noise dither during decoding. Could you please elaborate why "consistency" (meaning to obtain identical results when decoding through same tools from the same source) is bad? I was so sure until today, that SAME BINARY SOURCE DATA->SAME PROCESSING ALGORITHM->SAME BINARY OUTPUT in the world of computing...
The first law of audio processing is that when reducing bitdepth of digital audio data, you have to use dithering. Dithering means adding random noise to the data before rounding it down. The nature of "random" noise is that it's different every time. That's really all there is to it.

Ok, some algorithms use fixed noise patterns or calculate the noise the same way every time. With such algorithms you will get identical results if you repeat the same processing steps multiple times. But truely random noise is different every time. So being "inconsistent" is exactly the way to go - - but only when reducing audio bitdepth, of course.
madshi is offline  
Old 16th February 2009, 14:53   #8300  |  Link
hubblec4
Matroska find' ich toll
 
Join Date: Apr 2008
Posts: 1,370
Quote:
Originally Posted by madshi View Post
It's on my to do list, along with chapters and attachments.
wow.

you are the best one.


hubble
hubblec4 is offline  
Closed Thread

Tags
eac3to

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 23:03.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.