eac3to - audio conversion tool

[madshi]

Something is wrong with 1.2.1b (1.2.1a works fine though), but i forgot the problem (24 bit output broken?) :)
Ah - thanks!

[DreckSoft]

Aaaargh.
I've downloaded the file shortly after you posted the release of 2.08. It seems there was still the old zip online (or some browser cache problem). It was 2.07.

2.08 should work in both cases. Encoding is still in progress but it looks good.

Thanks!

[Snowknight26]

eac3to.exe "W:\Encoding Tools\temp\eragon.mpa" eragon.c2q.flac -down16 -libav
DTS Master Audio, 5.1 channels, 24 bits, 48khz
Decoding with DirectShow (Sonic Audio Decoder)...
Getting "Sonic Audio Decoder" instance failed.

Shouldn't be using Sonic, should it?

Edit: Ahh, nevermind, still need Sonic for DTS decoding it seems. I wonder if you'll ever implement DTS decoding using libdts or something similar.

[ACrowley]

eac3to.exe "W:\Encoding Tools\temp\eragon.mpa" eragon.c2q.flac -down16 -libav
DTS Master Audio, 5.1 channels, 24 bits, 48khz
Decoding with DirectShow (Sonic Audio Decoder)...
Getting "Sonic Audio Decoder" instance failed.

Shouldn't be using Sonic, should it?

Edit: Ahh, nevermind, still need Sonic for DTS decoding it seems. I wonder if you'll ever implement DTS decoding using libdts or something similar.

Why ? Sonic decodes DTS perfect. Including DTS HD and DTS ES 6.1
And its a Reference Decoder which should deliver better Quality compared to ffmpeg

[madshi]

Why ? Sonic decodes DTS perfect. Including DTS HD and DTS ES 6.1
And its a Reference Decoder which should deliver better Quality compared to ffmpeg
Exactly. Well, I could probably still add libav DTS decoding support. But I'm not sure if I really should do that since it can only be of worse quality compared to Sonic's decoder...

[Inventive Software]

Exactly. Well, I could probably still add libav DTS decoding support. But I'm not sure if I really should do that since it can only be of worse quality compared to Sonic's decoder...

Hold it... if it's a perfect implementation, it should be bit-identical to Sonic's version...

[madshi]

Hold it... if it's a perfect implementation, it should be bit-identical to Sonic's version...
But it isn't.

[Inventive Software]

What about using libdca, aka libdts?

[madshi]

Well, I'd have to spend precious time to support another totally new library without any real benefit - apart from not needing the Sonic decoder for conventional DTS tracks. And what about DTS-HD? No decoder apart from Sonic can decode DTS-HD! But if you want you can test that libdca/libdts decoder. If you find that it gives out results that are bit identical to Sonic for conventional DTS tracks then I might consider adding support for that.

[Inventive Software]

I actually don't use your software, but merely wanted to point out the other solutions. libdca's been around for a while, but is still in it's infancy. libdts is what's used in AC3Filter. And, I don't have Sonic's decoder so can't make the comparison! :(

[madshi]

I've compared AC3Filter's DTS decoding to Sonic's and Nero's DTS decoding. AC3Filter's decoding is too loud and definitely not bit identical.

Look, I appreciate suggestions (I really do!). It's just that I want to offer only the highest quality solutions. Furthermore my programming time is quite limited. Because of that I don't really want to spend time to add further decoders which are of worse quality than what eac3to already supports right now.

[nautilus7]

It's just that I want to offer only the highest quality solutions.That's the spirit!!! :D

[Inventive Software]

I've compared AC3Filter's DTS decoding to Sonic's and Nero's DTS decoding. AC3Filter's decoding is too loud and definitely not bit identical.

Look, I appreciate suggestions (I really do!). It's just that I want to offer only the highest quality solutions. Furthermore my programming time is quite limited. Because of that I don't really want to spend time to add further decoders which are of worse quality than what eac3to already supports right now.

I wondered what was wrong with Die Hard 4.0.... :D

I appreciate your lack of time, and didn't know libdts had deficiencies.

[madshi]

didn't know libdts had deficiencies.
Well, it's a general thing with open source lossy decoders (AC3 and DTS). I believe they're decoding "correctly", but the decoding code works differently than the reference decoder (used by Nero and Sonic) does. So the result is not bit identical. Since AC3 and (conventional) DTS are lossy codecs, it's probably not correct to say that the reference decoder is "right" and the open source decoders are "wrong". But I guess that the reference decoder should at least not be worse than the open source decoders, maybe slightly better. Using the reference decoders is the "safe" way IMHO. And when talking about DTS, Sonic is the only decoder that can decode DTS-HD. So IMO Sonic is absolutely the way to go for DTS decoding right now.

Not sure whether there's a problem with libdts or whether AC3Filter uses libdts incorrectly.

[maxpower2078]

I searched this post and couldn't seem to find this answer, so here I am.

I am trying to use this great program to extract the core DTS or AC3 tracks out of DTS-HD or DD+ tracks after demuxing them with EVOdemux and using the mpa file as the input.

with the following command:
eac3to sourcefile destfile -core

source is e-ac3, 5.1 ......640....

I get this operation is not permitted or something similar. Sorry, I am not in front of the machine right now.

I then tried to just downconvert it to see if that worked

eac3to sourcefile destfile -192


source is e-ac3..5.1....640....

I get the same message as well

I have the sonic 4.3 decoder installed as well as powerdvd ultra. I am wondering if I need the nero decoder or something?


My goal in this is to have the streams separate so that I can play then in media player classic and output over SPDIF in their full multi channel glory as I haven't been able to do so with the original format.

[nautilus7]

Only DTS-HD tracks have a simple DTS core in them. Extract it using:

eac3to input.dtshd output.dts -core

E-AC3 tracks don't have any AC3 core in them. Only Blu-ray TrueHD tracks might have AC3 frames (track) in them, but that's another thing.
To convert (encode) an E-AC3 track to AC3 you have to type:

eac3to input.eac3 output.ac3 -your desired bitrate

As you see in input file i set the extension to what is the real one. You don't have to do this. You can leave it as it is (.mpa). eac3to can automatically determine the input extension.

[maxpower2078]

Only DTS-HD tracks have a simple DTS core in them. Extract it using:

eac3to input.dtshd output.dts -core

E-AC3 tracks don't have any AC3 core in them. Only Blu-ray TrueHD tracks might have AC3 frames (track) in them, but that's another thing.
To convert (encode) an E-AC3 track to AC3 you have to type:

eac3to input.eac3 output.ac3 -your desired bitrate

As you see in input file i set the extension to what is the real one. You don't have to do this. You can leave it as it is (.mpa). eac3to can automatically determine the input extension.

Great, I'll give it a try when I get home, but the only thing different from your code is that I didn't put an .ac3 extension on my output file and kept the mpa extension too. Will this cause it to fail?

Also once I get past this step, how do I determine if there is a delay needed and what it would be?

Also, if I need to get the nero audio decoder, what version is the best to use? I know the changelog says to not use 8 and to use 7, but does it come in the bundle package or something?

[nautilus7]

Great, I'll give it a try when I get home, but the only thing different from your code is that I didn't put an .ac3 extension on my output file and kept the mpa extension too. Will this cause it to fail?
Sure it does!
.mpa doesn't mean anything to eac3to either it's input or output. It's just the extension that evodemux gives to all audio tracks (don't know where it comes from).
Input extension doesn't matter to be set correctly because, as i said before, eac3to can find out what is the input file by its structure, without have to look at the extension.
Output is a different thing. eac3to understands what has to do by looking to the output extension. If it's .ac3 understands that has to do ac3 encoding, if it's .flac understands that has to do flac encoding, etc.

Also once I get past this step, how do I determine if there is a delay needed and what it would be? You can get this value using evodemux. The track comes from an HD DVD, right? You have to load the first .evo of the movie and look at the Firts PTS value of the video and audio stream. There you can find the delay needed, if any.

Also, if I need to get the nero audio decoder, what version is the best to use? I know the changelog says to not use 8 and to use 7, but does it come in the bundle package or something?Only Nero 7 works and the decoders are included.

[maxpower2078]

Sure it does!
.mpa doesn't mean anything to eac3to either it's input or output. It's just the extension that evodemux gives to all audio tracks (don't know where it comes from).
Input extension doesn't matter to be set correctly because, as i said before, eac3to can find out what is the input file by its structure, without have to look at the extension.
Output is a different thing. eac3to understands what has to do by looking to the output extension. If it's .ac3 understands that has to do ac3 encoding, if it's .flac understands that has to do flac encoding, etc.


That is what I thought, ok, I'll try tonight.

The whole reason I am doing this as I said before was to be able to output the multichannel sound to a receiver via SPDIF which I can do with regular DVDs via ac3filter.

the HD-dvds are a little harder though. I am hoping if I can get these HD tracks down to good ol' AC3 or DTS I can mux them back together with the video and have it work via Media player classic or something.

The weird thing I ran into and haven't checked out yet is that the video .mpv elemental stream from the EVOdemux output plays with a pixalated green screen where the original plays just fine.

Has anyone run into this? Will it be find when I mux them back together?

[nautilus7]

Demuxing the video with evodemux doesn't always work.
The sasfest way for that is to use haali media splitter --> haali matroska muxer in graphedit, but it's not the right thread to discuss it.
Use search to find what you need. There are plenty of posts around.

[nautilus7]

@ madshi

Features request: I think it would be nice to add an option like press that key to cancel encoding. Also a message telling encoding took xxx seconds would be useful. These, if you have time and don't know how to spend it. :D

Finally, i updated the libaften.dll with the latest available (R703) and everything is working fine.

[madshi]

Only Nero 7 works and the decoders are included.
Just for completeness sake: Nero 7 is needed plus the Nero HD DVD / Blu-Ray plugin.

[madshi]

Features request: I think it would be nice to add an option like press that key to cancel encoding. Also a message telling encoding took xxx seconds would be useful. These, if you have time and don't know how to spend it. :D
As with every command line tool, you can press Ctrl+C to cancel encoding. That's Windows default behaviour for command line tools. Giving out encoding time is easy to add.

Finally, i updated the libaften.dll with the latest available (R703) and everything is working fine.
Good to know. According to jruggle the latest build should sound every so slightly better than the older builds. So I'll update to the latest libAften build with the next eac3to build.

[nautilus7]

As with every command line tool, you can press Ctrl+C to cancel encoding. That's Windows default behaviour for command line tools.Didn't know that. Thanks!

According to jruggle the latest build should sound every so slightly better than the older builds.
That's why i swapped it in the first place. :p Btw, i used the SSE3 enabled cause i have a core 2 duo.

[idbirch2]

Hi madshi, here's (http://forum.doom9.org/showthread.php?p=1072182#post1072182) the post I'm following up on, looks like quotes can't be quoted.

Currently eac3to's Aften encoder only supports 1, 2 and 6 channels. I'm not sure if Aften itself can encode 3/1 channels. Will have to check that.

It works for me with the sample you sent me. Maybe you can make a new sample which reproduces this new problem?

I have split the dtshd file into 10mb chunks with the hex editor you mentioned earlier but I still can't get it to work. Here's (http://rapidshare.com/files/74662017/esh001.dtshd.html) a 10MB chunk, if I try and convert this to wav, I just get:

F:\>"C:\Program Files\Audio\eAC3to\eac3to.exe" "H:\es\esh001.dtshd" "F:\esh1.wa
v"
DTS Master Audio, 3/1 channels, 24 bits, 48khz
Decoding with DirectShow (Sonic Audio Decoder)...
DirectShow reports 4 channels, 24 bits, 48khz
Writing WAV...
Creating/writing file "F:\esh1.wav"...
----------------------------------------------------------- How long should it take to do a 10MB chunk? So far I've been waiting 10 mins and it's not got any further.

[shambles]

i have a problem with a dts-hd master audio track.

DTS Master Audio, 5.1 channels, 24 bits, 48khz
Decoding with DirectShow (Sonic Audio Decoder)...
DirectShow reports 5.1 channels, 24 bits, 48khz
Encoding FLAC...
Creating/writing file "2.flac"...
This track is not clean. Processing aborted.
Please clean the track with delaycut and then retry eac3to.

it encodes fine until it gets to the very end. delaycut of course doesn't work with dts-hd tracks, but i tried trimming off a little bit from the end with HxD and saving the rest to a new file but then i get "The format of the source file could not be detected." i also tried running it through tsremux first and then demuxing but i get the same 'track is not clean' error.

when the processing is aborted, the output file is also deleted. is there any switch that would keep the output file? seeing as the error only occurs at the very end, i'm not sure there would be any actual audio data missing..

or better yet would be to fix the original file.. but how could that be done?

[nautilus7]

I had the exact same problem, but with an e-ac3 track. Delaycut couldn't correct it, so i used it just to get the frame that had the problem. Then i corrected it in a hex editor.
So, if you somehow find where exactly is the problem, you might be able to correct it in a hex editor. Can you upload the final part of the track? Try to include the erroneous part.

EDIT: That's a very good reason why madshi should update delaycut with dts-hd support.

[shambles]

with the split function in file tools in HxD i was able to produce a working dtshd file, and i believe i split it from exactly the spot where the bad frame is, seeing as now eac3to gets to the end fine with even the last frame being fully decoded (when trying to split it before, i got it wrong and eac3to would drop the last frame as it was incomplete)

http://rapidshare.com/files/74760031/broken.dtshd.html sample of the end of the track beginning with what i believe is the bad frame

[nautilus7]

Maybe i'm tottaly wrong, but it think the only problem with your sample is that the last frame isn't complete.
You can remove this frame by deleting the last 1455 (5AF in hex) bytes of the track. The sample is the end of the track, right?

@ madshi

Can you tell me the dts-hd frame structure/size please?

[shambles]

yes it's the end of the track. but if the last frame is incomplete, eac3to just says this "The last DTS frame is incomplete and thus gets skipped." and doesn't give any errors..

[G_M_C]

Am trying to get AC3 decoding to work with libav (doing a NTSC2PAL speedup, and demuxing into separate WAV's), but the -libav switch doesnt seem to work; EAC3To keeps wanting to use Nero, despite the swich).

(/me is not a fan of nero ....)

[nautilus7]

libav is only for e-ac3 and mpl/trueHD decoding.

[kakomu]

I've extracted an EAC3 track from a Matroska sample source using command line and mkvextract. I've also reinstalled Nero 7.9. When I used EAC3to with the EAC3 file, I receive this message when I try to encode into AC3:

E-AC3, 5.1 channels, 0:01:01, 640kbit/s, 48khz, dialnorm: -27dB
Removing dialog normalization...
Decoding with DirectShow (Nero Audio Decoder 2)...
Disabling DRC for Nero (E-)AC3 decoding...

When I try to encode into Wav, I get:

E-AC3, 5.1 channels, 0:01:01, 640kbit/s, 48khz, dialnorm: -27dB
Removing dialog normalization...
Decoding with DirectShow (Nero Audio Decoder 2)...
Disabling DRC for Nero (E-)AC3 decoding...
The WAV writer didn't receive the format information.

I must be missing something here. Is there some sort of guide on installation or am I SOL?

[nautilus7]

Do you have the HD DVD/Blu-ray plug-in for Nero?

[saint-francis]

I'm getting this error with every attempt to use eas3to now:

G:\mymovie>"C:\eac3to.exe" "G:\mymovie\featureXX.ddp" "G:\mymovie\featureXX.ac3" -640 -nero
source file "G:\mymovie\featureXX.ddp" not found.

I can assure you that the files are indeed there.

[nautilus7]

I think it has to do with the "" you put in the paths. Remove them to see what happens.

The -640 and -nero switches you specify are not needed because these are the parameters that eac3to uses by default.

[saint-francis]

This is just what eac3togui has given to eac3to.exe. I'll give it a shot from the command line though.

[kakomu]

Do you have the HD DVD/Blu-ray plug-in for Nero?

No, I do not. I take it that's required?

Is there a list of required software that is necessary to run this program successfully?

[nautilus7]

This is just what eac3togui has given to eac3to.exe. I'll give it a shot from the command line though.
You didn't tell it. Don't know cause i don't use the gui. Sorry.

[nautilus7]

No, I do not. I take it that's required?

Is there a list of required software that is necessary to run this program successfully?
Yes, it's required.

Have a look at the 1st post of this thread. I believe you 'll find what you need.

[The_Keymaker]

Hello fellow forum members,

I am in the process of upgrading EAC3toGUI to reflect the options in Madshi's latest version of eac3to.

In the meantime I am releasing an interim upgrade, v1.45. This does not have all the latest option switches, but it WILL let you type these option switches into the Command Line Preview box so they can be executed.

Remember to use the "Settings" menu to show EAC3toGUI where eac3to is located.

LINK: http://www.sendspace.com/file/9k2byp

I should have a full upgrade ready soon.

Thanks!
The_Keynmaker

[Thunderbolt8]

have some trouble with some audio pcm (?) files from the bruce springsteen - live at hammersmith odeon london '75 DVD. the DVD has 2 audio tracks, a 5.1 DTS track and a 2.0 LPCM track (the track of my choice)
this dvd is originally a video concert dvd, but dvd audio extractor lets me extract/convert the audio tracks to flac, wav or demux (pcm). I can choose the bitrate manually here in dvd audio extractor, but I dont know whether the original audio is 16- or 24-bit. thats why I tried the "demux" option and wanted to put those pcm files it gave me in eac3to, hoping it would run the 16 & 24-bit detection test and in the end remove possibly zero bites, if there were any.
but eac3to cant detect those types here so I made a sample. the 2 files are the 1st and 2nd track from that disc. the 1st (intro) being only ~15 seconds and from the 2nd track I cut a 10mb sample. the eac3to detection stops right at the beginning for the 2nd track, I guess thats because the track begins at once, without any audible break after the 1st track if they are in the playlist in that order.

http://www.sendspace.com/file/1bcba7

[saint-francis]

I'm getting this error with every attempt to use eas3to now:

G:\mymovie>"C:\eac3to.exe" "G:\mymovie\featureXX.ddp" "G:\mymovie\featureXX.ac3" -640 -nero
source file "G:\mymovie\featureXX.ddp" not found.

I can assure you that the files are indeed there.

The exact commands that the GUI gives work fine when I don't use the GUI. An issue with the GUI?

[madshi]

Here's a 10MB chunk, if I try and convert this to wav, I just get:

F:\>"C:\Program Files\Audio\eAC3to\eac3to.exe" "H:\es\esh001.dtshd" "F:\esh1.wa
v"
DTS Master Audio, 3/1 channels, 24 bits, 48khz
Decoding with DirectShow (Sonic Audio Decoder)...
DirectShow reports 4 channels, 24 bits, 48khz
Writing WAV...
Creating/writing file "F:\esh1.wav"...
----------------------------------------------------------- How long should it take to do a 10MB chunk? So far I've been waiting 10 mins and it's not got any further.
This sample decodes just fine for me (decoding takes about 5 seconds). Maybe your Sonic installation is not good? Try uninstalling and reinstalling Sonic.

[madshi]

with the split function in file tools in HxD i was able to produce a working dtshd file, and i believe i split it from exactly the spot where the bad frame is, seeing as now eac3to gets to the end fine with even the last frame being fully decoded (when trying to split it before, i got it wrong and eac3to would drop the last frame as it was incomplete)

http://rapidshare.com/files/74760031/broken.dtshd.html sample of the end of the track beginning with what i believe is the bad frame
Thanks. That was kind of a funny sample. The broken frame had one core block and two DTS-HD blocks. Basically the DTS-HD block was repeated twice. Very strange. The next build of eac3to will handle this situation. The Sonic decoder doesn't seem to care about such double DTS-HD blocks, so eac3to will allow that, too.

[madshi]

@ madshi

Can you tell me the dts-hd frame structure/size please?
Basically there's always a core frame and then a DTS-HD frame. The DTS-HD frame begins with "64 58 20 25". Directly after this DTS-HD "sign" there's the format indicator (High Resolution or Master Audio) and then the length of the DTS-HD block. The length is constant for High Resolution tracks. But for Master Audio every DTS-HD block can have a different length.

[madshi]

I'm getting this error with every attempt to use eas3to now:

G:\mymovie>"C:\eac3to.exe" "G:\mymovie\featureXX.ddp" "G:\mymovie\featureXX.ac3" -640 -nero
source file "G:\mymovie\featureXX.ddp" not found.

I can assure you that the files are indeed there.
Can't reproduce the problem here. Seems to work for me.

[madshi]

have some trouble with some audio pcm (?) files from the bruce springsteen - live at hammersmith odeon london '75 DVD. the DVD has 2 audio tracks, a 5.1 DTS track and a 2.0 LPCM track (the track of my choice)
this dvd is originally a video concert dvd, but dvd audio extractor lets me extract/convert the audio tracks to flac, wav or demux (pcm). I can choose the bitrate manually here in dvd audio extractor, but I dont know whether the original audio is 16- or 24-bit. thats why I tried the "demux" option and wanted to put those pcm files it gave me in eac3to, hoping it would run the 16 & 24-bit detection test and in the end remove possibly zero bites, if there were any.
but eac3to cant detect those types here so I made a sample. the 2 files are the 1st and 2nd track from that disc. the 1st (intro) being only ~15 seconds and from the 2nd track I cut a 10mb sample. the eac3to detection stops right at the beginning for the 2nd track, I guess thats because the track begins at once, without any audible break after the 1st track if they are in the playlist in that order.

http://www.sendspace.com/file/1bcba7
I'm not sure what these tracks are. They seem to be 16bit signed PCM. But converting them to either big or little endian both results in garbage. Maybe the demuxing didn't work correctly?

[nautilus7]

Basically there's always a core frame and then a DTS-HD frame. The DTS-HD frame begins with "64 58 20 25". Directly after this DTS-HD "sign" there's the format indicator (High Resolution or Master Audio) and then the length of the DTS-HD block. The length is constant for High Resolution tracks. But for Master Audio every DTS-HD block can have a different length.
The dts core is always 2013 bytes i guess.

I had a look at the sample and i saw that one frame was bigger than the others, but thought it was ok because it's a Master Audio track.
In that sample the last frame was ok? It's too sort i think.

[Thunderbolt8]

I'm not sure what these tracks are. They seem to be 16bit signed PCM. But converting them to either big or little endian both results in garbage. Maybe the demuxing didn't work correctly?
the option I selected was direct stream copy and he gave me these .pcm files for that. I also could have chosen output as waves, but eac3to wont alter bitdepth for these at flac conversion.