eac3to - audio conversion tool

[madshi]

I was curious as to why eac3to reported 47 chapters on a BD title (X-Files 2) when in fact standalone-player playback (and PowerDVD 8) offers only 27 - I started to poke around in BDInfo and BDedit.

BDInfo (0.5.2) reported the correct number of chapters (27) for playlist 00004.MPLS (playlist 00002.MPLS is a monolithic 1-chapter playlist of the same content by the looks of it, even though eac3to reports 47 chapters for it too)

Upon closer inspection, BDedit reveals that playlist 00004.MPLS has 46 of what it calls PlayListMarks. However, of those marks, only 27 are identified as having type "Entry-Mark" and these seem to tie up with actual chapter points, whereas the other PlayListMark types, which BDedit calls "Link Points", do not/should not feature in a chapter list.

Do you think this is right madshi? And can you/should you therefore exclude these "link point" types from your chapter lists??
Actually I'm already removing "link points" from the chapter list - if there is an "entry mark" somewhere afterwards. However, if there are "link points" after the last "entry mark" I'm not deleting these "link points". The reason for that is that at least one Blu-Ray had "entry marks" for the first half runtime of the movie and "link points" for the second half runtime of the movie.

Are those additional "link points" in X-Files 2 behind the last "entry point"? If not, I'd like to look into the problem. In that case please zip and upload the PLAYLIST and CLIPINF folders. Should only be a few KBs zipped.

[honai]

I guess I could add just another progress bar for this initial file scan. It usually causes a noticable delay only for seamless branching titles.

I think it might be enough to just output a string, "Performing initial file scan. This could take a few moments. Please wait ...", then continue as usual.

[magic144]

Hi madshi,
the link points seem to be interspersed with entry marks, so no, they're not after the last one.
Here's the files you wanted to check into the issue further. Hope this helps.

m

http://www.mediafire.com/?mnmxlytnyny

[laserfan]

For fun I just looked at Groundhog Day BD, and while it uses 16 chapters, all of eac3to, BDinfo, and BDedit show 31 (!) this latter having 31 entry-marks (only), no link-points ie. it's not branching at all.

So I don't know what's going on with some of the new titles. :confused:

[madshi]

the link points seem to be interspersed with entry marks, so no, they're not after the last one.
Here's the files you wanted to check into the issue further.
Actually with playlist 1, there is one entry mark, and then 46 link points. There's no further entry mark after all those link points. So that makes 47 chapters for playlist 1.

[magic144]

@madshi
sure, but it's playlist 4 that seems to be the right one (listed as the second title), which eac3to also lists as having 47 chapters

[magic144]

@laserfan
do none of the playlists for Groundhog Day register in BDinfo as having only 16 chapters then :-(
if so, there's apparently more variables to this riddle...

[laserfan]

do none of the playlists for Groundhog Day register in BDinfo as having only 16 chapters then...Good question! There are three .mpls I can find for the main movie (00001, 00220, 00221). Predictably (I guess, given I think madshi said he looked for longest .mpls?) eac3to finds 00221.mpls, which is the only one of the three with the incorrect chapter list! The other two do indeed show 16 chapters!

One has to wonder if authoring houses are doing this deliberately to confuse us re-doers and backer-uppers! :p

[Atak_Snajpera]

One has to wonder if authoring houses are doing this deliberately to confuse us re-doers and backer-uppers!
Most likely :) This would explain why main movie is very often divided in many parts without any particular order and size.

[madshi]

sure, but it's playlist 4 that seems to be the right one
And how is eac3to supposed to know that playlist 4 is the right one instead of playlist 1?

which eac3to also lists as having 47 chapters
There's a bug which result in eac3to always listing the default playlist's chapters, even if you manually specify a playlist. This will be fixed in the next build.

[mrr19121970]

I'm trying to demux the blu-ray 'While She Was Out'. This is what ea3to tells me about the stream:

eac3to v3.08
command line: "E:\TVIX\eac3to\eac3to.exe" "R:\" 1) 1: "D:\DEMUX\New\Chapters_1.txt" 2: "D:\DEMUX\New\Video_2.*" -slowdown 4: "D:\DEMUX\New\Audio_4_English.AC3" -slowdown 3: "D:\DEMUX\New\Audio_3_German.AC3" -slowdown -LOG="D:\DEMUX\New\eac3to_PASS3_LOG.LOG"
------------------------------------------------------------------------------
M2TS, 1 video track, 2 audio tracks, 1 subtitle track, 1:22:24, 50i
1: Chapters, 17 chapters
2: VC-1, 1080i50 (16:9)
3: DTS Master Audio, German, 5.1 channels, 16 bits, 48khz
(core: DTS, 5.1 channels, 16 bits, 1509kbps, 48khz)
4: DTS Master Audio, English, 5.1 channels, 16 bits, 48khz
(core: DTS, 5.1 channels, 16 bits, 1509kbps, 48khz)
5: Subtitle (PGS), German
Creating file "D:\DEMUX\New\Chapters_1.txt"...
[a04] The ArcSoft and Sonic decoders don't seem to work, will use libav instead.
[a04] The libav DTS decoder doesn't decode the full DTS-HD information. <WARNING>
[a03] The ArcSoft and Sonic decoders don't seem to work, will use libav instead.
[a03] The libav DTS decoder doesn't decode the full DTS-HD information. <WARNING>
[v02] Extracting video track number 2...
[v02] Writing new framerate "24fps /1.001" to bitstream.
[a03] Extracting audio track number 3...
[a03] Extracting DTS core...
[a03] Decoding with libav/ffmpeg...
[a03] Remapping channels...
[a03] Changing FPS from 25.000 to 23.976...
[a04] Extracting audio track number 4...
[a04] Extracting DTS core...
[a04] Decoding with libav/ffmpeg...
[a04] Remapping channels...
[a04] Changing FPS from 25.000 to 23.976...
[a03] Encoding AC3 <640kbps> with libAften...
[a04] Encoding AC3 <640kbps> with libAften...
[v02] Creating file "D:\DEMUX\New\Video_2.vc1"...
[a03] Creating file "D:\DEMUX\New\Audio_3_German.AC3"...
[a04] Creating file "D:\DEMUX\New\Audio_4_English.AC3"...
[a04] The original audio track has a constant bit depth of 64 bits.
[a04] The processed audio track has a constant bit depth of 64 bits.
[a03] The original audio track has a constant bit depth of 64 bits.
[a03] The processed audio track has a constant bit depth of 64 bits.
Video track 2 contains 123601 frames.
eac3to processing took 20 minutes, 43 seconds.
Done.


However the resulting VC1 stream is 48000/1001, even if I give '-stripPulldown' the resulting stream still has the interlaced flag. Is this normal ?


D:\DEMUX\While She Was Out _ 1>"E:\TVIX\eac3to\eac3to.exe" video_2.vc1
VC-1, 1080i48 /1.001 (16:9)

[madshi]

Just a couple of issues:
C:\unzipped\eac3to>eac3to.exe "C:\temp\i'm on a boat.mkv" 1: C:\temp\boat.h264 -
changeto29.970
MKV, 1 video track, 1 audio track, 0:03:09, 29.989p
1: h264/AVC, 720p29.989
2: AAC, 2.0 channels, 44.1khz
v01 The video bitstream is encoded in a non-standard framerate.
Was asked to modify track 1: to 29.970, but the original FPS is not supported.

C:\unzipped\eac3to>eac3to.exe "C:\temp\i'm on a boat.mkv" 1: C:\temp\boat.h264
MKV, 1 video track, 1 audio track, 0:03:09, 29.989p
1: h264/AVC, 720p29.989
2: AAC, 2.0 channels, 44.1khz
v01 The video bitstream is encoded in a non-standard framerate.
v01 Extracting video track number 1...
v01 Creating file "C:\temp\boat.h264"...
v01 Video overlaps for 1 frames at playtime 0:00:00.
v01 Video overlaps for 1 frames at playtime 0:00:01.
v01 Video overlaps for 1 frames at playtime 0:00:01.
v01 Video overlaps for 1 frames at playtime 0:00:02.
[...]
v01 Video overlaps for 1 frames at playtime 0:03:09.
v01 Video overlaps for 1 frames at playtime 0:03:09.
v01 Video overlaps for 1 frames at playtime 0:03:10.
eac3to processing took 10 seconds.
Done.
These problems will be fixed in the next build. The problems were caused by the really strange framerate, which is really "29.989fps". Even the video bitstream contains exactly this framerate.

I'm trying to demux the blu-ray 'While She Was Out'. This is what ea3to tells me about the stream:

eac3to v3.08
command line: "E:\TVIX\eac3to\eac3to.exe" "R:\" 1) 1: "D:\DEMUX\New\Chapters_1.txt" 2: "D:\DEMUX\New\Video_2.*" -slowdown 4: "D:\DEMUX\New\Audio_4_English.AC3" -slowdown 3: "D:\DEMUX\New\Audio_3_German.AC3" -slowdown -LOG="D:\DEMUX\New\eac3to_PASS3_LOG.LOG"
------------------------------------------------------------------------------
M2TS, 1 video track, 2 audio tracks, 1 subtitle track, 1:22:24, 50i
1: Chapters, 17 chapters
2: VC-1, 1080i50 (16:9)
3: DTS Master Audio, German, 5.1 channels, 16 bits, 48khz
(core: DTS, 5.1 channels, 16 bits, 1509kbps, 48khz)
4: DTS Master Audio, English, 5.1 channels, 16 bits, 48khz
(core: DTS, 5.1 channels, 16 bits, 1509kbps, 48khz)
5: Subtitle (PGS), German
Creating file "D:\DEMUX\New\Chapters_1.txt"...
[a04] The ArcSoft and Sonic decoders don't seem to work, will use libav instead.
[a04] The libav DTS decoder doesn't decode the full DTS-HD information. <WARNING>
[a03] The ArcSoft and Sonic decoders don't seem to work, will use libav instead.
[a03] The libav DTS decoder doesn't decode the full DTS-HD information. <WARNING>
[v02] Extracting video track number 2...
[v02] Writing new framerate "24fps /1.001" to bitstream.
[a03] Extracting audio track number 3...
[a03] Extracting DTS core...
[a03] Decoding with libav/ffmpeg...
[a03] Remapping channels...
[a03] Changing FPS from 25.000 to 23.976...
[a04] Extracting audio track number 4...
[a04] Extracting DTS core...
[a04] Decoding with libav/ffmpeg...
[a04] Remapping channels...
[a04] Changing FPS from 25.000 to 23.976...
[a03] Encoding AC3 <640kbps> with libAften...
[a04] Encoding AC3 <640kbps> with libAften...
[v02] Creating file "D:\DEMUX\New\Video_2.vc1"...
[a03] Creating file "D:\DEMUX\New\Audio_3_German.AC3"...
[a04] Creating file "D:\DEMUX\New\Audio_4_English.AC3"...
[a04] The original audio track has a constant bit depth of 64 bits.
[a04] The processed audio track has a constant bit depth of 64 bits.
[a03] The original audio track has a constant bit depth of 64 bits.
[a03] The processed audio track has a constant bit depth of 64 bits.
Video track 2 contains 123601 frames.
eac3to processing took 20 minutes, 43 seconds.
Done.

However the resulting VC1 stream is 48000/1001, even if I give '-stripPulldown' the resulting stream still has the interlaced flag. Is this normal ?
eac3to can not make truely interlaced content progressive. Such a thing is not possible without fully reencoding the video track. eac3to can only remove the pulldown from an HD DVD style video track. Such HD DVD style tracks are progressively encoded, but contain additional pulldown flags. The Blu-Ray you have seems to be native interlaced. No way to fix that. However, playback of "1080i48 /1.001" should work fairly well, too.

AFAIK Blu-Ray does not support 1080p25 content. If you want PAL framerate, you have to use 1080i50. Don't ask me why the Blu-Ray companies have made such a stupid decision...

[madshi]

eac3to v3.09 released

http://madshi.net/eac3to.zip

* added support for MKV "SRT/UTF8", "SRT/ASCII", "ASS" and "SSA" subtitles
* increased some internal buffers to avoid AC3 overflow in the "thd ac3 joiner"
* fixed: frame counting didn't work for MKV video tracks
* fixed: video tracks FPS change was sometimes declined
* fixed: video tracks with "strange" FPS were sometimes handled incorrectly
* fixed: clipping removal 2nd pass was executed even for "stdout"
* fixed: "eac3to -test" displayed an outdated Nero download link
* fixed: specifying a specific playlist still used default playlist's chapters

[Atak_Snajpera]

AFAIK Blu-Ray does not support 1080p25 content. If you want PAL framerate, you have to use 1080i50. Don't ask me why the Blu-Ray companies have made such a stupid decision...
Officially you are right. However I believe that players which have AVCHD label should play those files without any problems. There are HD-camcorder like Canon HF10/100 (AVCHD structure) which can record in 25p. PS3 proves my theory and plays without any complaints :)

[mrr19121970]

the 1080i50 is a shop bought blu-ray. i guess it's tsMuxeR that's wrecking it, but thankfully a straight copy of the main stream onto my TVIX HDD is playable.

tsMuxeR GUI doesn't even reckognise the VC1 stream, and the CLI accepts it and chews it up. I'll try again with manually setting the framerate to see if that helps.

[Snowknight26]

Thanks madshi! The youtube video works now.

One thing I noticed, though, is that the language for subtitles is no longer displayed.

[magic144]

And how is eac3to supposed to know that playlist 4 is the right one instead of playlist 1?

There's a bug which result in eac3to always listing the default playlist's chapters, even if you manually specify a playlist. This will be fixed in the next build.

cheers for the fix madshi - just tested 3.09 on X-Files 2 and it now lists and extracts 27 chapters for title 2 - thanks!!

as to identifying which title/playlist is the "right" title, I wouldn't know how to figure that out programatically, I guess the easiest way for now is just manually, i.e. to try the disc in a standalone player or software player and see what the chapters are there and then see which title eac3to identifies that matches

presumably, to do this programatically, you'd have to do what a player does and navigate through the disc's playback logic, taking into account user menu choices and such-like along the way - no doubt far more than eac3to was ever envisioned as doing! - I think the way it is right now is perfectly fine!!

[laserfan]

eac3to v3.09 released

* fixed: specifying a specific playlist still used default playlist's chaptersI just tried this on Groundhog Day, and although it still selects the 00221.mpls playlist with (incorrect) 31 chapters, I can now direct it to 00001.mpls with (correct) 16 chapters shown!

Thanks madshi for the update! :)

[itsancho]

oops, no languages with 3.09 C:\Users\iTSAN>z\eac3to "H:\Lost (2004) (TV series) The Complete Fourth Season
2007) Blu-ray AVC PCM\Disc 1" 2)
M2TS, 1 video track, 7 audio tracks, 22 subtitle tracks, 0:43:09, 24p /1.001
1: Chapters, 8 chapters
2: h264/AVC, 1080p24 /1.001 (16:9)
3: AC3, 5.1 channels, 640kbps, 48khz
4: RAW/PCM, 5.1 channels, 16 bits, 48khz
5: DTS, 5.1 channels, 24 bits, 1509kbps, 48khz
6: AC3, 5.1 channels, 640kbps, 48khz
7: DTS, 5.1 channels, 24 bits, 1509kbps, 48khz
8: AC3, 5.1 channels, 640kbps, 48khz
9: AC3, 2.0 channels, 192kbps, 48khz, dialnorm: -30dB
10: Subtitle (PGS)
11: Subtitle (PGS)
12: Subtitle (PGS)
13: Subtitle (PGS)
14: Subtitle (PGS)
15: Subtitle (PGS)
16: Subtitle (PGS)
17: Subtitle (PGS)
18: Subtitle (PGS)
19: Subtitle (PGS)
20: Subtitle (PGS)
21: Subtitle (PGS)
22: Subtitle (PGS)
23: Subtitle (PGS)
24: Subtitle (PGS)
25: Subtitle (PGS)
26: Subtitle (PGS)
27: Subtitle (PGS)
28: Subtitle (PGS)
29: Subtitle (PGS)
30: Subtitle (PGS)
31: Subtitle (PGS)

[Thunderbolt8]

* fixed: frame counting didn't work for MKV video tracksdoes this refer to the frame adding of the video file after remuxing or was this only a display bug? would we have to redo some files because of that if fluent playback is possibly not given?

[DrNein]

The Nero Audio Decoder works just fine for me in v3.08.

Hmm, this is odd. I retested and it is "not working correctly" in versions 3.06 through 3.09 but with earlier versions it "works fine". The problem is not just in reporting either as it actually falls back to libav/ffmpeg.

[madshi]

does this refer to the frame adding of the video file after remuxing or was this only a display bug?
Just a display bug.

Hmm, this is odd. I retested and it is "not working correctly" in versions 3.06 through 3.09 but with earlier versions it "works fine". The problem is not just in reporting either as it actually falls back to libav/ffmpeg.
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.

[madshi]

eac3to v3.10 released

http://madshi.net/eac3to.zip

* Blu-Ray title listing now includes chapter information
* fixed: v3.09 didn't show track languages for Blu-Rays

[Snowknight26]

* fixed: v3.09 didn't show track languages for Blu-Rays

Still doesn't show them for mkv files.

[madshi]

Still doesn't show them for mkv files.
Does for me. Does the MKV file you're testing actually have language information in it? :)

[Snowknight26]

Microsoft Windows [Version 6.0.6001]
Copyright (c) 2006 Microsoft Corporation. All rights reserved.

C:\unzipped\eac3to>eac3to.exe "Z:\Movies\Chungking Express\Chung Hing sam lam.mkv" -logmkv
EBML
DocTypeId: "matroska"
DocTypeVersion: 1
DocTypeReadVersion: 1
Segment
Meta Seek Information
Seek
SeekID: (binary, len: 4)
SeekPosition: 4099
Seek
SeekID: (binary, len: 4)
SeekPosition: 4249
Seek
SeekID: (binary, len: 4)
SeekPosition: 13110584910
Seek
SeekID: (binary, len: 4)
SeekPosition: 13110552269
Void: (binary, len: 4025)
Segment Information
TimecodeScale: 1000000
MuxingApp: "libebml v0.7.7 + libmatroska v0.8.1"
WritingApp: "mkvmerge v2.4.2 ('Oh My God') built on Jan 18 2009 17:30:28"
Duration: 1:42:56.160
DateUTC: 2/2/2009 3:48:18 AM
SegmentUID: (binary, len: 16)
Track
Track Entry
TrackNumber: 1
TrackUID: 1
TrackType: video
FlagEnabled: 1
FlagDefault: 1
FlagForced: 0
FlagLacing: 0
MinCache: 1
TrackTimecodeScale: 1
MaxBlockAdditionID: 0
CodecID: "V_MPEG4/ISO/AVC"
CodecDecodeAll: 1
CodecPrivate: (binary, len: 41)
DefaultDuration: 41708299
Language: "chi"
Video
PixelWidth: 1800
PixelHeight: 1080
FlagInterlaced: 0
DisplayWidth: 5
DisplayHeight: 3
Track Entry
TrackNumber: 2
TrackUID: 707616952
TrackType: audio
FlagEnabled: 1
FlagDefault: 1
FlagForced: 0
FlagLacing: 1
MinCache: 0
TrackTimecodeScale: 1
MaxBlockAdditionID: 0
CodecID: "A_DTS"
CodecDecodeAll: 1
Language: "chi"
Audio
SamplingFrequency: 48000
Channels: 6
Track Entry
TrackNumber: 3
TrackUID: 4182481851
TrackType: subtitle
FlagEnabled: 1
FlagDefault: 1
FlagForced: 0
FlagLacing: 0
MinCache: 0
TrackTimecodeScale: 1
MaxBlockAdditionID: 0
CodecID: "S_TEXT/ASS"
CodecDecodeAll: 1
CodecPrivate: (binary, len: 912)
Language: "eng"
Void: (binary, len: 1024)
Cluster
MKV, 1 video track, 1 audio track, 1 subtitle track, 1:42:56, 24p /1.001
1: h264/AVC, Chinese, 1800x1080 23.976p
2: DTS, Chinese, 5.1 channels, 24 bits, 1509kbps, 48khz
3: Subtitle (ASS)
v01 The video bitstream is encoded in a non-standard framerate.

C:\unzipped\eac3to>

It does.

[itsancho]

1st is 3.10, 2nd - 3.08 C:\Users\iTSAN>z\eac3to "L:\Firefly (2002) (TV series 2002-2003) The Complete Se
ries Blu-ray AVC dts-HD MA [107]\13 Heart of Gold.mkv"
MKV, 1 video track, 1 audio track, 2 subtitle tracks, 0:42:28, 24p /1.001
1: h264/AVC, English, 1080p24 /1.001 (16:9)
"AVC 18.9"
2: DTS Master Audio, English, 5.1 channels, 24 bits, 48khz
(core: DTS, 5.1 channels, 24 bits, 1509kbps, 48khz)
"dts-HD MA 5.1 24-Bit 3648"
3: Subtitle (SRT)
4: Subtitle (SRT)

C:\Users\iTSAN>zz\eac3to "L:\Firefly (2002) (TV series 2002-2003) The Complete S
eries Blu-ray AVC dts-HD MA [107]\13 Heart of Gold.mkv"
MKV, 1 video track, 1 audio track, 2 subtitle tracks, 0:42:28, 24p /1.001
1: h264/AVC, English, 1080p24 /1.001 (16:9)
"AVC 18.9"
2: DTS Master Audio, English, 5.1 channels, 24 bits, 48khz
(core: DTS, 5.1 channels, 24 bits, 1509kbps, 48khz)
"dts-HD MA 5.1 24-Bit 3648"
3: TEXT/UTF8, Bulgarian
4: TEXT/UTF8, English
Bitstream parsing for tracks 3 and 4 failed.
Demuxing these tracks may still produce correct results - or not.

[madshi]

eac3to.exe "Z:\Movies\Chungking Express\Chung Hing sam lam.mkv" -logmkv
Ouch. Proven wrong by my own debug switch... :o

Video and audio track languages did work, but subtitle track languages not.

[madshi]

eac3to v3.11

http://madshi.net/eac3to.zip

* fixed: MKV subtitle track language wasn't shown

<sigh> Hopefully this will be the last build for this week. Please don't find any more bugs, guys... :p

[itsancho]

thank u, madshi! great support, quick response! thanks again!

[DrNein]

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.

[73ChargerFan]

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.

[odin24]

Does eac3to still detect forced subs on BDs? If so it didn't on Iron Man (North American).

Thanks,

[Snowknight26]

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.

[kypec]

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 (http://www.mediafire.com/?sharekey=8b5e96c28ff29ad7e62ea590dc5e5dbbe04e75f6e8ebb871), 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) T:\eac3to\eac3to.exe "test.VOB" "R:\test.wav" -down16 -libav
Compatible WAV output (old style PCM header)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.
http://preview.shareapic.net/preview4/014695611.png (http://www.shareapic.net/content.php?id=14695611&owner=kypec) http://preview.shareapic.net/preview4/014695617.png (http://www.shareapic.net/content.php?id=14695617&owner=kypec)
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.:confused:
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...:eek:

[madshi]

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.

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.

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.

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...:eek:
There are no weird issues. Your testing method is invalid.

[gigah72]

the languages are missing

3.08

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

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)

[T800]

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. :)

[kypec]

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.

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:
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 (http://www.mediafire.com/file/ayjwgngye1k/5tests.7z). Each of these 5 files is UNIQUE!:mad:
Can anybody else confirm eac3to produces consistent output with WAV files (using built-in libav only). You can use my test.VOB file (http://www.mediafire.com/?sharekey=8b5e96c28ff29ad7e62ea590dc5e5dbbe04e75f6e8ebb871) 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...:thanks:

[honai]

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. :)

[kypec]

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.:confused:

[madshi]

the languages are missing

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).

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.

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.

[buzzqw]

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

[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
Yes, of course. You have full control over what gets demuxed, transcoded, whatever. E.g. you can do:

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.

[hubblec4]

@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

[buzzqw]

yup ! thansk madshi

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

BHH

[madshi]

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.

[kypec]

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:rolleyes:

[madshi]

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.

[hubblec4]

It's on my to do list, along with chapters and attachments.

wow.

you are the best one.


hubble