eac3to - audio conversion tool

[tebasuna51]

here is a sample of mine ac3 i'll cut off small part with delaycut 1.3
all my ac3 from dvb cap. remuxed with haali splitter
same resulting file with nero splitter
http://rapidshare.com/files/90978719/Flushed_Away_fixed.ac3

delaycut - report:

You need post the DelayCut log because detect the problem:
your file begin with 5 frames (166 ms) stereo but at frame 6 is 5.1 until the end of sample. This kind of mixes are typical from tv captures and you need split the ac3 before decode.

[nautilus7]

when playing MPC report that the file is ac3 6 channel....

...maybe it a problem with a header in ac3 or eac3to not correct detect this type file
I believe it's a 5.1 ac3, but for some reason eac3to and delaycut detect it as 2.0.

Madshi should have a look into it.

EDIT: tabasuna51 caught me!

[K-rnivoro]

E-AC3 decoding is still limited to 5.1 But now the libav decoder doesn't crash and can extract the 5.1 core.

So, it is not be able to decode 6.1 or 7.1 files to 7 or 8 wavs? I got it all wrong.....

[menlvd]

You need post the DelayCut log because detect the problem:
your file begin with 5 frames (166 ms) stereo but at frame 6 is 5.1 until the end of sample. This kind of mixes are typical from tv captures and you need split the ac3 before decode.

delaycut log: ac3 demuxed from ts (splited with haali media splitter)
[Input info]
Bitrate=384
Actual rate=384.000000
Sampling Frec=48000
TotalFrames=152059
Bytesperframe=1536.0000
Filesize=233562624
FrameDuration= 32.0000
Framespersecond= 31.2500
Duration=01:21:05.888
Channels mode=2/0: L+R
LFE=LFE: Not present
[Target info]
StartFrame=0
EndFrame=1388
NotFixedDelay= 0.0000
Duration=00:00:44.448
====== PROCESSING LOG ======================
Time 00:00:00.160; Frame#= 6. Some basic parameters changed between Frame #1 and this frame
Number of written frames = 1389
Number of Errors= 1

[Snowknight26]

I'm trying to back up my FF7AC DVD but I found some problems:


eac3to.exe VTS_01_1.VOB+VTS_01_2.VOB+VTS_01_3.VOB+VTS_01_4.VO
B+VTS_01_5.VOB+VTS_01_6.VOB+VTS_03_1.VOB
The file "VTS_01_2.VOB" doesn't look like a valid EVO/VOB file.
VOB, 1 video track, 2 audio tracks
1: Joined VOB file
2: MPEG2, 480p24 /1.001
3: AC3, 5.1 channels, 448kbit/s, 48khz, dialnorm: -25dB
4: AC3, 5.1 channels, 448kbit/s, 48khz, dialnorm: -25dB

eac3to.exe VTS_01_2.VOB
This doesn't seem to be a valid (E-)AC3 file.
This doesn't seem to be a valid (E-)AC3 file.
VOB, 1 video track
1: MPEG2, 480p24 /1.001

clip.exe
usage: clip <infile> <outfile> <start offset> <length>

clip VTS_01_2.VOB VTS_01_2_CUT.VOB 0 20000000

eac3to.exe VTS_01_2_CUT.VOB
This doesn't seem to be a valid (E-)AC3 file.
This doesn't seem to be a valid (E-)AC3 file.
VOB, 1 video track
1: MPEG2, 480p24 /1.001


I can play the file just fine (the sample too) but eac3to doesn't like it.

Here is the sample: http://www.stfcc.org/misc/VTS_01_2_CUT.VOB

Edit: Before, I couldn't even extract the audio tracks, but now I can. Strange. Also, if I add the -nero switch, it seems to work fine too, even though I don't have Nero installed and eac3to says so when I use -test.

[Thunderbolt8]

got a problem here, just found out that mkvmerge apparently cannot delay flac tracks (at least not properly) and since I did that instead of using eac3to for quite a while I might redo a lot of my movies. however, there are some which didnt need any delay at all and if I knew to which of them this did apply I could save myself a lot of time, because then I didnt have to re-ripp the disc just to check the .evos for a delay.
the possible candidates are:

2001: a space odyssey
300
a clockwork orange
army of shadows (french studio canal)
eyes wide shut
la haine (french studio canal)
rambo first blood
serpico
the graduate (german arthaus edition)
the shining


does someone maybe know some of these titles which DEFINATELY not need a delay? would help me a lot, thanks!

[jmrenrav]

just curios, try it please with v2.20 or v2.19, which you can get here: http://www.afterdawn.com/software/au...ools/ac3to.cfm

Link doesn't work, its the same with 2.20 but i cant find a download link for 2.19 =p

[Thunderbolt8]

strange, it worked for me back then a few hours :S

anyway, here is 2.19 again: http://www.sendspace.com/file/r31iew

[Yraen]

got a problem here, just found out that mkvmerge apparently cannot delay flac tracks (at least not properly) and since I did that instead of using eac3to for quite a while I might redo a lot of my movies. however, there are some which didnt need any delay at all and if I knew to which of them this did apply I could save myself a lot of time, because then I didnt have to re-ripp the disc just to check the .evos for a delay.


Could you not just use eac3to file1.evo+file2.evo from the drive without ripping them? This would show you the delay, if any.

For what it's worth, Matrix only shows a delay of -3ms on track 12 (E-AC3, 2 channel audio), none of the others

[Thunderbolt8]

ok then I can already cancel matrix from the list, thanks!

[umaximus]

got a problem here, just found out that mkvmerge apparently cannot delay flac tracks (at least not properly) and since I did that instead of using eac3to for quite a while I might redo a lot of my movies. however, there are some which didnt need any delay at all and if I knew to which of them this did apply I could save myself a lot of time, because then I didnt have to re-ripp the disc just to check the .evos for a delay.
the possible candidates are:

2001: a space odyssey
300
a clockwork orange
army of shadows (french studio canal)
batman begins
eyes wide shut
la haine (french studio canal)
letters from iwo jima
mulholland drive (french studio canal)
rambo first blood
serpico
terminator 2 extended edition (french studio canal)
the departed
the graduate (german arthaus edition)
the pianist (french studio canal)
the shining
v for vendetta

does someone maybe know some of these titles which DEFINATELY not need a delay? would help me a lot, thanks!

batman begins +200ms/trueHD
letters from iwo jima - no delay/trueHD
mulholland drive - no delay/dts-hd
T2 DirCut - no delay/dts-hd
the departed - no delay/trueHD
the pianist - no delay/dts-hd
v for vendetta - no delay/trueHD

[tebasuna51]

delaycut log: ac3 demuxed from ts (splited with haali media splitter)
...
Time 00:00:00.160; Frame#= 6. Some basic parameters changed between Frame #1 and this frame
...

When DelayCut send this message the ac3 fixed is unusable for most decoders. In this case is a change from 2.0 -> 5.1 but the same message is send if change the samplerate (44.1 -> 48 KHz) or the bitrate.

[Thunderbolt8]

batman begins +200ms/trueHD
letters from iwo jima - no delay/trueHD
mulholland drive - no delay/dts-hd
T2 DirCut - no delay/dts-hd
the departed - no delay/trueHD
the pianist - no delay/dts-hd
v for vendetta - no delay/trueHD
thanks, helped me a big part so far!

[The_Keymaker]

Please download the latest version (v1.85) of EAC3toGUI from the following link:

http://www.sendspace.com/file/s193u2

Version 1.85 changes and upgrades:
1. Now supports two source Files for joining EVO/VOB files. Default extension for joined output file is the same as main (first) source file.
2. More pre-conversion error checking to ensure source and destination files are of the correct type.
3. Added *.vc1 as a valid input File.
4. Deleted -ignoreGaps option (removed in eac3to v2.22).

I have not exhaustively tested the file joining feature, so please provide feedback on if it works.

I will work on adding more options and enhancements soon. In the meantime you can always use the "Other Options" tab to input options manually.

Please report any bugs or suggestions for improvements.

Regards,
The_Keymaker

[Thunderbolt8]

a quick question, im sure it has already been answered here, but I cant find it :S
when I demux a truehd or dts-hd track which has a delay e.g. theres -14ms in the name, when I later on want to convert it to flac only from this file and not from the main .evos any more, will I have to apply this -14ms then myself manually, or is this just information that this track needs -14ms delay, but eac3to will still add that itself?

[Chumbo]

a quick question, im sure it has already been answered here, but I cant find it :S
when I demux a truehd or dts-hd track which has a delay e.g. theres -14ms in the name, when I later on want to convert it to flac only from this file and not from the main .evos any more, will I have to apply this -14ms then myself manually, or is this just information that this track needs -14ms delay, but eac3to will still add that itself?
I went back and forth with madshi about automatically applying delay, i.e., I didn't want it done automatically. ;) So if you're using eac3to, then it probably is applying the delay automatically. If you look at the status text that is displayed, you should see something like "applying delay" to the audio track. At least that's the case when going to dts or ac3. I don't use flac.

[Beastie Boy]

eac3to cannot apply delay to TrueHD or DTS-HD tracks. If a file is written with a delay in the file name, then this delay will need to be manually applied when you convert to another format.

Cheers, Beastie.

[Thunderbolt8]

is there a switch in eac3to which disables adding a specific delay? for example, for rebuilt .evo with audio only in it from evodemux theres another delay specified as the actual track has (i guess because the reference to the video beginning timestamps are missing), so e.g. a track really needs -83ms (reported by eac3to when examining the .evos), but eac3to gives +42ms for this rebuilt .evo. eac3to automatically applies this value then, even though i'd be -83ms whats needed. so do I have to make a 2nd step each time now by adding the difference of those 2 values like adding -125ms then to my flac to get to my -83ms?

[nautilus7]

If eac3to reports +42ms for a track, you can specify -42ms manually. The result will be +42ms -42ms = 0ms.

[Thunderbolt8]

will this manually -42ms or whatever I type there then override the automatic and wrong +42ms from eachto so that I would really end up with -42ms? or would it only cancel out each other, and i'd have 0 in the end?

[nautilus7]

They would cancel each other and you will end up with 0ms.

[Thunderbolt8]

ok thanks!

[shambles]

i finally have my new box together.. with my ati 3850, in mpc-hc i get full dxva if i remux a vc1 hd dvd with gdsmux but only green/purple/white corruption if i remux with eac3to. the eac3to remux works fine in software mode tho

h.264 remuxing seems to work fine

[rickardk]

Do you use Haali Media Splitter or MPC:s own splitter?

[shambles]

both have same result

[umaximus]

madshi, new version is realy picky about delaycut thing. I have encounterd several movies that v.2.22 didnt want to work, but everything is fine with v.2.20. Any idea why this?

[Thunderbolt8]

ye, I have the same problems with some tracks and especially with older rebuilt .evos which contain only the truehd or dtshd track of a movie, as I already wrote in #3393

[rickardk]

i finally have my new box together.. with my ati 3850, in mpc-hc i get full dxva if i remux a vc1 hd dvd with gdsmux but only green/purple/white corruption if i remux with eac3to. the eac3to remux works fine in software mode tho

h.264 remuxing seems to work fine

Codec used?
eac3to version?

I can't reproduce this with my 8600GTS.
Will try 3870 tonight.

[shambles]

newest eac3to and mpc-hc's internal vc1 decoder which only supports full bitstream dxva which the nvidia cards don't do

[bmnot]

New eac3to version is smokin fast!
To remux a 20.2 GB file to mkv with v2.20 it took 57 minutes.
To remux it with v2.22 it took 14 minutes!!!

Wow!
Thanks!

[madshi]

btw. has eac3to become more strict regarding the acceptance of some audio tracks over time? back then i rebuilt the dts-hd ma track of pans labyrinth as seperate evo with evodemux and now I tried to make a flac out of it again but i get the same "not clean" error as above. but back then it worked :S
I'm having a problem with a audio track from X-Men 3 Bluray. I think its a DTS-ES audio track. I'm getting the "not clean" error message AFTER i run it though the cut program, otherwsie it says the "source is not reconized" any optopns?
madshi, new version is realy picky about delaycut thing. I have encounterd several movies that v.2.22 didnt want to work, but everything is fine with v.2.20. Any idea why this?
Sorry guys, this is a bug introduced in the latest eac3to version, caused by the change which I did to support shorter DTS files. Will be fixed in the next version.

eac3to cannot apply delay to TrueHD or DTS-HD tracks. If a file is written with a delay in the file name, then this delay will need to be manually applied when you convert to another format.
eac3to can apply delay to DTS-HD tracks but not to TrueHD tracks.

newest eac3to and mpc-hc's internal vc1 decoder which only supports full bitstream dxva which the nvidia cards don't do
This is most probably a bug in the MPC HC's internal VC-1 decoder. eac3to muxes have the full VC-1 stream info in it, while Haali's EVO/m2ts splitter removes a lot of stuff from the VC-1 stream. I guess that the MPC HD's internal VC-1 decoder is written specifically to handle Haali's VC-1 output. Which is not good. All other VC-1 decoders (Microsoft, Sonic, ffdshow) accept eac3to MKV remuxes just fine - actually they work better with eac3to's remuxes. So this is something that needs to be fixed in the MPC HC's internal VC-1 decoder.

I'm trying to back up my FF7AC DVD but I found some problems:

eac3to.exe VTS_01_2.VOB
This doesn't seem to be a valid (E-)AC3 file.
This doesn't seem to be a valid (E-)AC3 file.
VOB, 1 video track
1: MPEG2, 480p24 /1.001

I can play the file just fine (the sample too) but eac3to doesn't like it.
It seems that the AC3 stream in the VTS_01_2.VOB does not begin at the start of an AC3 frame. Instead the VOB begins in the middle of an AC3 frame. eac3to doesn't like this (at all). I'm not even sure if it's allowed for a VOB to behave like this. Anyway, you could try joining the VOB files via "copy /b" and then run them through eac3to. You might get a wrong runtime shown by eac3to that way (or not), but demuxing/remuxing should actually work.

New eac3to version is smokin fast!
To remux a 20.2 GB file to mkv with v2.20 it took 57 minutes.
To remux it with v2.22 it took 14 minutes!!!

Wow!
:)

[madshi]

Please download the latest version (v1.85) of EAC3toGUI from the following link:

http://www.sendspace.com/file/s193u2

Version 1.85 changes and upgrades:
1. Now supports two source Files for joining EVO/VOB files. Default extension for joined output file is the same as main (first) source file.
2. More pre-conversion error checking to ensure source and destination files are of the correct type.
3. Added *.vc1 as a valid input File.
4. Deleted -ignoreGaps option (removed in eac3to v2.22).

I have not exhaustively tested the file joining feature, so please provide feedback on if it works.

I will work on adding more options and enhancements soon. In the meantime you can always use the "Other Options" tab to input options manually.
Thanks! I've uploaded it to my homepage, as usual.

[shambles]

This is most probably a bug in the MPC HC's internal VC-1 decoder. eac3to muxes have the full VC-1 stream info in it, while Haali's EVO/m2ts splitter removes a lot of stuff from the VC-1 stream. I guess that the MPC HD's internal VC-1 decoder is written specifically to handle Haali's VC-1 output. Which is not good. All other VC-1 decoders (Microsoft, Sonic, ffdshow) accept eac3to MKV remuxes just fine - actually they work better with eac3to's remuxes. So this is something that needs to be fixed in the MPC HC's internal VC-1 decoder.

vc1 in evo and m2ts works too in mpc-hc tho..

[madshi]

vc1 in evo and m2ts works too in mpc-hc tho..
By using the Haali splitter, I guess? Try the Sonic splitter instead. You'll probably see the same problems you see with the eac3to muxes with it.

[Thunderbolt8]

eac3to can apply delay to DTS-HD tracks but not to TrueHD tracks.
so when I demux a dts-hd track from a movie to store it for later purpose is this then already done automatically by eac3to, if that track had a delay, or would I have to do this myself? and can delaying dts-hd tracks be done up to 1ms precise as can flac tracks?

[madshi]

so when I demux a dts-hd track from a movie to store it for later purpose is this then already done automatically by eac3to, if that track had a delay, or would I have to do this myself? and can delaying dts-hd tracks be done up to 1ms precise as can flac tracks?
eac3to delays every track it can (which is all except TrueHD/MLP). For those where delaying doesn't work, eac3to is writing the missing delay into the file name. So you never need to worry. Either the delay is automatically done by eac3to, or it is written to the file name.

DTS-HD delaying is similar to (E-)AC3 delaying: It's not possible to do it 1ms precise without reencoding. However, if you let eac3to convert the EVO->DTS-HD track directly to FLAC (which is the default action for DTS-HD Master Audio tracks when doing "eac3to source.evo dest.mkv"), the delay is done with 1ms precision.

[nautilus7]

The delay in compressed tracks is frame length precise. Is eac3 frame length standard (5,33ms) or it depends on the bitrate? What is the case in DTS-hd tracks?

[Thunderbolt8]

eac3to delays every track it can (which is all except TrueHD/MLP). For those where delaying doesn't work, eac3to is writing the missing delay into the file name. So you never need to worry. Either the delay is automatically done by eac3to, or it is written to the file name.

DTS-HD delaying is similar to (E-)AC3 delaying: It's not possible to do it 1ms precise without reencoding. However, if you let eac3to convert the EVO->DTS-HD track directly to FLAC (which is the default action for DTS-HD Master Audio tracks when doing "eac3to source.evo dest.mkv"), the delay is done with 1ms precision.
hm so then basically when I demux a dts-hd track the delay will be approximately right, but not exactly? what if I convert this track to flac afterwards then, is the very correct information still stored somewhere that the flac will be 1ms precise afterwards again or will the flac only be that approx. correct as the dts-hd track is?

not sure, wasnst eac3 frame length the same as ac3 with 32ms? and dts-hd was 10.666ms? so this means the demuxed dts-hd track is max. 10.666ms off?

[madshi]

hm so then basically when I demux a dts-hd track the delay will be approximately right, but not exactly? what if I convert this track to flac afterwards then, is the very correct information still stored somewhere that the flac will be 1ms precise afterwards again or will the flac only be that approx. correct as the dts-hd track is?
If you demux first, the delay is not precise and will never be. But we're talking about a few milliseconds only. I don't think it matters. delaycut can only delay unprecise and nobody ever complained about that, as far as I remember...

Is eac3 frame length standard (5,33ms) or it depends on the bitrate? What is the case in DTS-hd tracks?
not sure, wasnst eac3 frame length the same as ac3 with 32ms? and dts-hd was 10.666ms? so this means the demuxed dts-hd track is max. 10.666ms off?
To be honest, I don't know right now.

[nautilus7]

not sure, wasnst eac3 frame length the same as ac3 with 32ms? and dts-hd was 10.666ms? so this means the demuxed dts-hd track is max. 10.666ms off?

No, definitely not. Eac3 frame length is not that long. Only ac3 is 32ms. I 've seen eac3s with a frame length of 5,33ms only. Load a eac3 track in delaycut to see yourself. I don't have a clue about dts-hd though.

Regarding max value of delay that a track could be off...
I think it is only half a frame length, or at least that's the way it should be. For example (in case of ac3 - 32ms frame length) if you request to delay that track by 1 - 15 ms then no delay should be applied (1 - 15ms is rounded down to 0ms). If you want to delay by 16 -47ms then 32ms (1 frame) is added, and so on.

But that's the worst case as ac3 has the longer frame. For eac3 5,33ms frame length the error is minimum.

[tebasuna51]

The delay in compressed tracks is frame length precise. Is eac3 frame length standard (5,33ms) or it depends on the bitrate? What is the case in DTS-hd tracks?

In ac3 the frame timelength is fix (32 ms. for 48 KHz) because always have 6 blocks (256 samples per block) per frame.

In eac3 a frame can have 1, 2, 3 or 6 blocks (256, 512, 768 or 1536 samples) and 5.33, 10.67, 16 or 32 ms. if 48 KHz.

When uncompressing a format we obtain samples then any delay can be sample length precise:
1/48000 sec = 1/48 ms. = 0.020833 ms. (also for flac)
but is useless something less than 1 ms.

[nautilus7]

In ac3 the frame timelength is fix (32 ms. for 48 KHz) because always have 6 blocks (256 samples per block) per frame.

In eac3 a frame can have 1, 2, 3 or 6 blocks (256, 512, 768 or 1536 samples) and 5.33, 10.67, 16 or 32 ms. if 48 KHz.

When uncompressing a format we obtain samples then any delay can be sample length precise:
1/48000 sec = 1/48 ms. = 0.020833 ms. (also for flac)
but is useless something less than 1 ms.
Thanks!
Do blocks/samples have any relation with bitrate?

[tebasuna51]

Do blocks/samples have any relation with bitrate?

Yes but not direct.
In eac3 header not exist the bitrate value, there are:
- SampleRate (16, 22.05, 24, 32, 44.1 or 48 KHz)
- FrameLength (any even value until 4096)
- Number of Blocks per Frame (1, 2, 3 or 6)

We can obtain the same bitrate with:
FrameLength = 4096 and Blocks per Frame = 6 (32 ms/frame)
FrameLength = 2048 and Blocks per Frame = 3 (16 ms/frame)

[nautilus7]

Yes but not direct.
In eac3 header not exist the bitrate value, there are:
- SampleRate (16, 22.05, 24, 32, 44.1 or 48 KHz)
- FrameLength (any even value until 4096)
- Number of Blocks per Frame (1, 2, 3 or 6)

We can obtain the same bitrate with:
FrameLength = 4096 and Blocks per Frame = 6 (32 ms/frame)
FrameLength = 2048 and Blocks per Frame = 3 (16 ms/frame)

I see. Do you know where i can find detailed info about eac3 frame header? Thanks again.

[tebasuna51]

I see. Do you know where i can find detailed info about eac3 frame header? Thanks again.
About E-AC3 you can read the Anex E of Digital Audio Compression Standard (AC-3, E-AC-3) Revision B (http://www.atsc.org/standards/a_52b.pdf)

[shambles]

By using the Haali splitter, I guess? Try the Sonic splitter instead. You'll probably see the same problems you see with the eac3to muxes with it.

you're right, doesn't work with sonic either

[madshi]

you're right, doesn't work with sonic either
Casimir666 is aware of the problem and will probably fix it in a future MPC HD build.

[Joniii]

How does eac3to detect if channels need to be remapped? I mean I ripped TrueHD from Blu-ray the usual way but this time eac3to didn't say anything about this being Blu-ray TrueHD and didn't remap channels:

D:\2.22>eac3to e:\audio.thd d:\audio.dts
TrueHD/AC3, 5.1 channels, 48khz
Extracting TrueHD stream...
Writing WAVs...
Creating/writing file "d:\audio.L.wav"...
Creating/writing file "d:\audio.C.wav"...
Creating/writing file "d:\audio.SL.wav"...
Creating/writing file "d:\audio.R.wav"...
Creating/writing file "d:\audio.SR.wav"...
Creating/writing file "d:\audio.LFE.wav"...
Found Surcode DTS Encoder version 1.0.23.0.
Surcode encoding successfully started. Please wait...

[madshi]

5.1 TrueHD (both Blu-Ray and HD DVD) decodes to the normal WAV channel order, so channels are already mapped correctly.

[Joniii]

5.1 TrueHD (both Blu-Ray and HD DVD) decodes to the normal WAV channel order, so channels are already mapped correctly.
dang, didn't remember that. thx.