eac3to - audio conversion tool

[madshi]

The 4 bytes beforehand are actually part of the same frame - the first two bytes indicate the frame length, the next two are a form of DTS - these are then followed by the sync word. Without the bytes, 128 packets of 40 samples will be dropped - about 107ms by my reckoning.
That's interesting - thanks for the information!

Is it possible to get the runtime from those DTS bytes? I mean could I just search in the file for the first and last sync word, substract the DTS values and convert them to a runtime somehow? That would be nice.

If you've got any samples with extra junk on the front I'd like to get a look at them - PM me if that's possible.
Right now I don't have any at hand. But I'm sure I'll run across them again sooner or later. Will PM you then.

[shambles]

it seems weird that .raw is not a valid input extension since eac3to decodes to just that. renaming it to .pcm works but then eac3to seems to assume it's a blu-ray lpcm track and wants to remap the channels.

i wanted to use -down6 on a raw file that was decoded from flac, but i just couldn't get it to work..

'eac3to bla.pcm bla.flac -8 -down6' makes a 6ch flac but channels are mapped wrong

'eac3to bla.pcm bla.flac -8 -0,1,2,3,4,5,6,7 -down6' makes an 8ch flac

flac to flac with -down6 doesn't work either.. 'eac3to bla.flac bla2.flac -down6' creates an 8ch file

it of course works just fine with the orig blu-ray lpcm track but not anymore once the channels have been remapped.

also .wav input seems to be broken.. even wavs made by eac3to give the error message "The raw file has 0 channels, which is not supported by eac3to."

[madshi]

it seems weird that .raw is not a valid input extension since eac3to decodes to just that. renaming it to .pcm works but then eac3to seems to assume it's a blu-ray lpcm track and wants to remap the channels.

i wanted to use -down6 on a raw file that was decoded from flac, but i just couldn't get it to work..

'eac3to bla.pcm bla.flac -8 -down6' makes a 6ch flac but channels are mapped wrong

'eac3to bla.pcm bla.flac -8 -0,1,2,3,4,5,6,7 -down6' makes an 8ch flac

flac to flac with -down6 doesn't work either.. 'eac3to bla.flac bla2.flac -down6' creates an 8ch file

it of course works just fine with the orig blu-ray lpcm track but not anymore once the channels have been remapped.

also .wav input seems to be broken.. even wavs made by eac3to give the error message "The raw file has 0 channels, which is not supported by eac3to."
Probably those are all bugs. And probably they are all already fixed with eac3to v2.0.

[madshi]

eac3to v2.0 released

http://madshi.net/eac3to.zip

Please see first post of this thread for the full help text of the new version. Here I'll post a list of the changes instead:

totally new features
* AC3 decoding support (Nero's decoder without DRC/dialnorm)
* resampling to 44.1/48/96 kHz (by using "r8brain")
* apply/reverse PAL speedup (by using "r8brain")
* "eac3to sourceFile" will print out source file details

strongly enhanced features
* dramatically improved performance (no intermediate files, anymore!)
* proper 6.1/7.1 downmixing to 5.1 instead of just dropping the back surround channels
* RAW/PCM file detection now auto detects channels, bitdepth and endian
* WAV is now fully supported as source file format
* destination file extension "PCM" creates Blu-Ray style LPCM tracks
* bitdepth can be reduced to anything between 14 bits and 23 bits

DTS related improvements/changes
* DTS-96/24 support added
* "open bitrate" support added
* strange channel configuration support added
* removal of zero padding from DTS files added
* eac3to can fix broken DTS-ES files (they decode to 5.1 instead of 6.1 without the fix)
* dialog normalization can be removed without removing the additional DTS-HD data now
* core extraction must be specifically asked for now (see "-core" switch)

AC3 related improvements
* did I mention that eac3to can decode AC3 now?
* strange channel configuration support added

TrueHD related improvements
* delay problem (hopefully) solved
* fixed: sometimes some audio data in the middle of a track was lost
* TrueHD/AC3 interweaved file can be stripped to TrueHD only now

various minor improvements/changes
* progress bar added
* eac3to detects file format independently of file extension
* multiple input files can be treated as one big file
* "sox" is not needed, anymore
* "dump" filter not needed, anymore
* "aften.exe" replaced by "libAften.dll"
* "flac.exe" replaced by "libFlac.dll"
* DTS/DD+/AC3 source filter ships with eac3to now
* 8bit support added
* various bug fixes
Probably I've forgotten to mention some changes. There's one thing I should note, though: This is a total rewrite! That means, there may be tons of new bugs. So we'll need to do a lot of testing before we can consider it really stable. One thing I'm not sure about yet is how well the new version will work in Vista, especially in Vista64. I've implemented a number of ugly hacks and tested them only in XP. So eventually I'll need to fix some things in Vista.

Just to be sure, here's the previous eac3to v1.23 for download:

http://madshi.net/eac3to.v1.23.zip

[Thunderbolt8]

:thanks:

edit: hows the quality of PAL speedup removal compared to an original NTSC track, is it of same quality? when slowing down to NTSC with besweet for example there was still some noticeable difference

[laserfan]

Thanks madshi, I just started playing w/eac3to today, and here pops your post! Like the sourcefile printout, and already noticed it prints my "Nero missing" error instantly this time! ;)

I see that Nero 7 is dirt-cheap, and the plugins are $25; but I have not used Nero for years as it's bloated with stuff I neither need nor want on my computer.

Has anyone here figured-out how to minimally install the audio decoder from Nero 7, and the Plugins without all that other junk, er, good stuff?

[Thunderbolt8]

i will remux letters from iwo jima again later tonight, will be able to say then if that truehd delay thing is fixed now

btw. could you please explain what this means:
* destination file extension "PCM" creates Blu-Ray style LPCM tracks ?

when I want to convert blue-ray movie LPCM tracks to flac, do I still have to rename the tracks to .pcm before I put them into eac3to, or to .lpcm or doesnt it matter which of both?

[madshi]

hows the quality of PAL speedup removal compared to an original NTSC track, is it of same quality? when slowing down to NTSC with besweet for example there was still some noticeable difference
Just try it out and report back! :) Quality should be a bit better than besweet. But the original NTSC track is of course hard to beat.

[madshi]

Has anyone here figured-out how to minimally install the audio decoder from Nero 7, and the Plugins without all that other junk, er, good stuff?
Not sure. But the installer of Nero 7 Essentials should allow you to choose which packages to install. Maybe you can even disable "everything". Haven't tried that yet...

[madshi]

btw. could you please explain what this means:
* destination file extension "PCM" creates Blu-Ray style LPCM tracks ?
That means that you can e.g. do "eac3to source.ac3 dest.pcm" and you'll get a PCM file which is formatted the same way as Blu-Ray PCM files are. Not sure if it's useful for anything, but there were some people who were asking for that. So...

when I want to convert blue-ray movie LPCM tracks to flac, do I still have to rename the tracks to .pcm before I put them into eac3to, or to .lpcm or doesnt it matter which of both?
The new RAW/PCM detection algorithm costs some time. So I'm using it only if the file extension is "raw" or if there's a "PCM" somewhere in the file name. Not sure but IIRC xport puts a "PCM" in the file name? If it does, you can just convert the file without having to rename it. Otherwise rename it to "*.pcm" or "*.raw". That's not too hard - is it!?

[Thunderbolt8]

xport puts .mpa at the end.

[madshi]

xport puts .mpa at the end.
I said "if there's a "PCM" somewhere in the file name".

[Thunderbolt8]

you didnt get my question at all, I asked if have to rename the file to .pcm or .lpcm for blue-ray flac tracks, since the the source input option now is (L)PCM and not only .pcm as in previous versions.
found out it actually doesnt matter.

[nautilus7]

I have to make an announcement: MADSHI IS A GREAT LIER!!!

...and i explain:
1. He was telling that new eac3to won't be ready till next week,
2. that he hadn't make any progress to AC3 decoding,
3. that PAL<--> NTSC conversion won't be ready for the next version.

3 big lies :D:D:D


Thank you very much!!!

:thanks::thanks::thanks:

[madshi]

I have to make an announcement: MADSHI IS A GREAT LIER!!!

...and i explain:
1. He was telling that new eac3to won't be ready till next week,
2. that he hadn't make any progress to AC3 decoding,
3. that PAL<--> NTSC conversion won't be ready for the next version.

3 big lies :D:D:D


Thank you very much!!!
Sorry for lying... :D

No honestly, I didn't really expect AC3 decoding and PAL<->NTSC in the first version. The reason why AC3 decoding made it into this version is very simple: I noticed that my E-AC3 Nero related hacks worked for AC3, too, without any changes. So it was really easy to add AC3 decoding.

[nautilus7]

I am currently installing windows on my server... After that i 'll do some conversions. Get ready for feedback.

[Thunderbolt8]

remuxed letters from iwo jima again, both audio and video had same first timestamps. therefore I didnt apply any delay to the truehd flac and everything is in sync. seems to work perfertly now, thanks again!

[nautilus7]

I found out some more features that you didn't tell.
A very good bug report. Yeah, i had a crash. It's the same file i mentioned a few post above. Can you tell anything about this?

[shambles]

fantastic. i don't know how i could live without this tool and now it's a tenfold better..

you're my hero.

by the way, i'm guessing noise-shaping is used with the TPDF dithering? any idea how the quality is compared to ssrc/wavewizard? really interested to see what kind of results 24->20bit conversion will bring.. :D

[nautilus7]

C:\Tools>eac3to ice.greek.ac3 ice.greek.new.ac3 -slowdown -resampleto48000
AC3, 5.1 channels, 1:26:53, 384kbit/s, 48khz
Creating/writing file "ice.greek.new.ac3"...
Done.

Tried something different this time.
The process finished almost in no time and the output file is identical to the source.
Did i do something wrong?

[madshi]

I found out some more features that you didn't tell.
A very good bug report. Yeah, i had a crash. It's the same file i mentioned a few post above. Can you tell anything about this?
Not sure. Can you please mail that bug report to me? A sample of that file would help, too (if the crash also occurs with the sample).

[madshi]

by the way, i'm guessing noise-shaping is used with the TPDF dithering?
No noise shaping. It's "simple" TPDF dithering. Should be good enough, as long as you don't go under 16bit. Noise shaping only reduces the noise that you hear. I don't think TPDF noise is audible in 16bit.

any idea how the quality is compared to ssrc/wavewizard?
No idea. The best idea would be: Try it out and report it here!

[madshi]

C:\Tools>eac3to ice.greek.ac3 ice.greek.new.ac3 -slowdown -resampleto48000
AC3, 5.1 channels, 1:26:53, 384kbit/s, 48khz
Creating/writing file "ice.greek.new.ac3"...
Done.

Tried something different this time.
The process finished almost in no time and the output file is identical to the source.
Did i do something wrong?
Ah, that's a bug. Here's what eac3to thinks:

"Source file is ac3, destination file is ac3, and no different bitrate was specified. So I'm not supposed to reencode. I'm only supposed to remove dialnorm (if even necessary)."

You can work around that by doing it in 2 steps:

(1) eac3to ice.greek.ac3 ice.greek.wav -slowdown
(2) eac3to ice.greek.wav ice.greek.new.ac3

Please note that the "-resampleto48000" is probably superfluous in your case. AC3 files coming from DVDs are usually already 48000. And "-slowdown" does not change the samplerate.

[nautilus7]

I 'll send you the bug report, but how? I don't use an email client. Only webmail (gmail). And a sample too.

For the 2nd problem, i 'll do what you suggested.

[madshi]

really interested to see what kind of results 24->20bit conversion will bring.. :D
Ah yes, that's one thing I forgot to mention: Bitdepth can be reduced to anything between 14 bits and 23 bits now. And for FLAC encoding every bit helps getting the file size down. Roger Dressler (Dolby) recently stated on AVSForum that in his opinion 20 bits may be the sweet spot. So personally I might now reduce all my 24 bit tracks to 20 bit. Haven't fully decided yet...

[hristoff2]

You certainly should re-think the name 'eac3to' :D

We're so reliant on you and your work, thanks a million. :)

[Mtz]

Can somebody post a command line for an eac3 source and the desired target to be AC3 640kbps using the sonic filter?
This is not working anymore for me:
eac3to stream1.eac3 stream2.ac3 -48 -640 -sonic

enjoy,
Mtz

[madshi]

I 'll send you the bug report, but how? I don't use an email client. Only webmail (gmail). And a sample too.
You can PM me the bug report or you can post it into gmail. Once the bug report window is visible on screen, the bug report is automatically stored into the clipboard. So just open PM or gmail and press Ctrl+V.

[madshi]

You certainly should re-think the name 'eac3to' :D
Yeah, that's true. I was already thinking about alternative names. But the name "eac3to" is well known now, even if it doesn't really fit, anymore. So I'm not sure if I should change the name...

[madshi]

Can somebody post a command line for an eac3 source and the desired target to be AC3 640kbps using the sonic filter?
This is not working anymore for me:
eac3to stream1.eac3 stream2.ac3 -48 -640 -sonic
What is that "-48" supposed to mean? Just remove that. It should work then. The "-640" is probably not necessary, either, cause it's the default value used by eac3to for multichannel audio tracks. So just "eac3to stream1.eac3 stream2.ac3 -sonic" should do.

[hristoff2]

Yeah, that's true. I was already thinking about alternative names. But the name "eac3to" is well known now, even if it doesn't really fit, anymore. So I'm not sure if I should change the name...

Indeed it's well known, you've got a point there. Guess that's the same reason why 'avimux' is still named avimux although it supports matroska etc etc. :)

[Mtz]

Still not working. Error log sent via PM. The problem appear after: Encoding AC3...
Sonic decoder version 4.2.

enjoy,
Mtz

[Thunderbolt8]

I have a problem with ratatouille remuxing (AVC, LPCM).
the movie has seamless branching so I joined each .m2ts part via "copy /b..". after that I used xport to extract video and audio from the joined .m2ts file and used eac3to to convert the LPCM stream (24-bit) to FLAC (everything fine).
but after I remuxed both together with mkvmerge I get an audio problem at playback (ffdshow, coreflac): audio suddenly disappeares at ~53 seconds movietime and also doesnt come back later on, shortly after that mpc crashes.
im not quite sure whether this could be a problem of the new eac3to version or because of seamless branching joining; however the error occurs before the first switch from the 1st to 2nd .m2ts file took place (length of 1st .m2ts file ~1:30 mins).

im currently using v1.23 to convert to flac again and will mux it all again together to see whether it might still crash then.

[madshi]

I have a problem with ratatouille remuxing (AVC, LPCM).
the movie has seamless branching so I joined each .m2ts part via "copy /b..". after that I used xport to extract video and audio from the joined .m2ts file and used eac3to to convert the LPCM stream (24-bit) to FLAC (everything fine).
but after I remuxed both together with mkvmerge I get an audio problem at playback (ffdshow, coreflac): audio suddenly disappeares at ~53 seconds movietime and also doesnt come back later on, shortly after that mpc crashes.
im not quite sure whether this could be a problem of the new eac3to version or because of seamless branching joining; however the error occurs before the first switch from the 1st to 2nd .m2ts file took place (length of 1st .m2ts file ~1:30 mins).

im currently using v1.23 to convert to flac again and will mux it all again together to see whether it might still crash then.
What happens if you play the external FLAC file? Do you get audio after those 53 seconds? Does madFlac also fail to work? Why are you using ffdshow/coreflac? madFlac should be better...

Usually after having joined TS files you need to run TsRemux over the result. However, I'm not sure if that does anything to audio. Maybe as a result of the seamless branching the joined audio data is corrupt. Would be strange but not impossible. E.g. if you add one byte in the middle of a valid PCM file you'll get totally trash sound after that byte.

[nautilus7]

You can PM me the bug report or you can post it into gmail. Once the bug report window is visible on screen, the bug report is automatically stored into the clipboard. So just open PM or gmail and press Ctrl+V.OK, i got it.

I also think a rename of eac3to is needed.

[madshi]

Still not working. Error log sent via PM. The problem appear after: Encoding AC3...
Sonic decoder version 4.2.
It seems that libAften is crashing for you. It's probably a multithreaded related bug. I'll switch libAften.dll to single threading in the next build, just to be safe.

[madshi]

New eac3to name suggestions:

- madAvConvert(er) - (AV for AudioVideo)
- madHcConvert(er) - (Hc for HomeCinema)
- madConvert(er)

The AudioVideo just in case. If I name it "AudioConverter" now, I'll have to change again in case I should ever decide to add some kind of video processing.

Your opinions? Any other suggestions?

[Thunderbolt8]

What happens if you play the external FLAC file? Do you get audio after those 53 seconds? Does madFlac also fail to work? Why are you using ffdshow/coreflac? madFlac should be better...

Usually after having joined TS files you need to run TsRemux over the result. However, I'm not sure if that does anything to audio. Maybe as a result of the seamless branching the joined audio data is corrupt. Would be strange but not impossible. E.g. if you add one byte in the middle of a valid PCM file you'll get totally trash sound after that byte.
afaik madflac was only for external flac files, but my file was inside the .mkv
I tried to mux the flac file into .mka and wanted to play it outside, but I got an error directly before at the beginning of the parsing stage (and strangely enough this error was not reported or didnt stop the muxing of the previous .mkv with video and audio together).

[madshi]

afaik madflac was only for external flac files, but my file was inside the .mkv
I tried to mux the flac file into .mka and wanted to play it outside, but I got an error directly before at the beginning of the parsing stage (and strangely enough this error was not reported or didnt stop the muxing of the previous .mkv with video and audio together).
The newer versions of madFlac support both external files and also files inside of MKV or other containers.

[Thunderbolt8]

The newer versions of madFlac support both external files and also files inside of MKV or other containers.
tried to use madflac instead of coreflac for the remuxed audio/video file, but it gave this grey filter screen directly at the beginning and no sound filter at all was loaded then.

the mkvmerge remux with the v1.23 flac file just finished parsing successfully, will soon find out whether it will work together with the video.

[nautilus7]

New eac3to name suggestions:

- madAvConvert(er) - (AV for AudioVideo)
- madHcConvert(er) - (Hc for HomeCinema)
- madConvert(er)

The AudioVideo just in case. If I name it "AudioConverter" now, I'll have to change again in case I should ever decide to add some kind of video processing.

Your opinions? Any other suggestions?I like the fisrt the most, but not very very much. I 'll propose one later.

The ctrl+V doesn't work. It doesn't copy the error report. Do you want only the first tab of it?
I found also that the error report window is always in front of the other windows, like task manager. It's little annoying.


EDIT: I found out the log.txt in the eac3to folder, so no problem.

[Thunderbolt8]

the audio is fine when using eac3to v1.23 (apart from some delay, cause by the way of joining the files together I guess), the overall filesize is also bigger, because of the successful flac parsing.

the filesize of both tracks is also different, depending on the use of which version:

FLAC v.123: 2.77 GB
FLAC v2.0: 2.73 GB

I made a 10mb sample of the 3 tracks (source .pcm, flac v1.23, flac v2.0), maybe it helps.
http://www.megaupload.com/de/?d=09KGU6AN

[nautilus7]

Ah, that's a bug. Here's what eac3to thinks:

"Source file is ac3, destination file is ac3, and no different bitrate was specified. So I'm not supposed to reencode. I'm only supposed to remove dialnorm (if even necessary)."

You can work around that by doing it in 2 steps:

(1) eac3to ice.greek.ac3 ice.greek.wav -slowdown
(2) eac3to ice.greek.wav ice.greek.new.ac3

Please note that the "-resampleto48000" is probably superfluous in your case. AC3 files coming from DVDs are usually already 48000. And "-slowdown" does not change the samplerate.It just came to me...

The output wav would exceeds the 4GB limit. It's a 1,5 hour movie. So, is there any other method?

[nautilus7]

About the eac3 to ac3 conversion i am trying to do... Here's the log and a sample, but i think it's a libaften related problem, because eac3 to wavs works.

http://rapidshare.com/files/70467275/sample.zip.html

[Thunderbolt8]

the audio is fine when using eac3to v1.23 (apart from some delay, cause by the way of joining the files together I guess), the overall filesize is also bigger, because of the successful flac parsing.

the filesize of both tracks is also different, depending on the use of which version:

FLAC v.123: 2.77 GB
FLAC v2.0: 2.73 GB

I made a 10mb sample of the 3 tracks (source .pcm, flac v1.23, flac v2.0), maybe it helps.
http://www.megaupload.com/de/?d=09KGU6AN
did another remux try, the first attempt was only with copy /b and then xport use. did now run tsremux over the with copy /b merged .m2ts before using xport, but the result was the same: audio still out of sync in the big .mkv during the movie (flac with v1.23) and the FLAC file created with v20 still cant be put into .mka, again parsing error direct at the beginning. might really be a little bug of the new version :P

[ACrowley]

@Madshi excellent work

I like the AC3 decoding Feature with a Reference Decoder (nero) so the AC3 to X Quality should be optimal.
Also the Timestrech is very nice..no need for me to use Behappy anymore..now i will use eac3to for "ALL" Codecs/Formats to .wavs

Hopefull Keymaker create a new gui because gui 1.5 fails from the beginning on...Filepath in ,", quotes will not work with eac3to 2.0 etc etc etc

[Sephiroth0000]

Hiya eveeryone! Madshi loving the new EAC3TO. Although on that note when I attempted to make a DTS file from it does not finish up. It does all the WAV but yet at the end it saids something like SURCODE DTS ENCODER DOES'NT SEEM TO BE INSTALLED

Any idea's anyone?

[menlvd]

Hiya eveeryone! Madshi loving the new EAC3TO. Although on that note when I attempted to make a DTS file from it does not finish up. It does all the WAV but yet at the end it saids something like SURCODE DTS ENCODER DOES'NT SEEM TO BE INSTALLED

Any idea's anyone?

have same problem! but in my sys installed 1.0.29 ver of SurCode DVD-DTS

[Sephiroth0000]

I did not even know I was supposed to install anything else. What else am I supposed to get to make it work?

[madshi]

have same problem! but in my sys installed 1.0.29 ver of SurCode DVD-DTS
Please check your registry. If Surcode is properly installed for you there should be a registry value named:

HKEY_LOCAL_MACHINE\Software\Minnetonka Audio Software\SurCode DVD DTS\Home

eac3to is checking if that value exists. If it doesn't exist eac3to complains that Surcode is not installed.