eac3to - audio conversion tool

[LigH]

I preferred to mix the LFE with -3 dB when I used Azid.

[robertcollier4]

After many downmixes to play AC3 5.1 movies on my stereo hardware player, here are the settings I have found optimal. A batch file included if it might be useful to others.

My findings:
1) I prefer the Nero 7 AC3 decoder (if you have it installed), thus I force it over ffmpeg using the -nero switch. I have not had any problems with the "bug in disabling Nero DRC" in a few hundred encodes.
2) I prefer -downStereo over -downDpl because it gives cleaner dialogues. Although -downDpl effects might sound a little better due to the 'matrix' echo effect in left channel, I found dialogues to be cleaner in -downStereo.
3) I prefer -mixlfe as it includes bass that gets picked up by the crossover on my 2.1 speaker set.
4) I use -full to let eac3to pass qaac 64 bit-depth data since that is what is generated by the eac3to downmixer anyways. No need to do a down dithering operation since qaac will take care of that when it converts the audio data to floating point.
5) I use qaac because it has --no-delay which removes the 44ms encoder delay introduced by all AAC encoders. Thus the audio is insured to be in sync with the original video.

@echo off
set PATH_EAC3TO=D:\PortableApps\VideoProcessing\eac3to327
set PATH_MKVMERGE=D:\PortableApps\VideoProcessing\mkvtoolnix-unicode-6.0.0
set PATH_QAAC=D:\PortableApps\VideoProcessing\qaac
set PATH=%PATH%;%PATH_EAC3TO%;%PATH_MKVMERGE%;%PATH_QAAC%

for /f %%a IN ('dir /b *.mkv') do (
echo ---------------------- BATCH STARTING %%~na%%~xa
eac3to.exe "%%~na%%~xa" -log="%%~na.eac3to.log" 2: stdout.wav -full -nero -downStereo -mixlfe -normalize | qaac.exe --tvbr 127 --quality 2 --rate keep --ignorelength --no-delay - -o "%%~na.m4a" 1>"%%~na.log" 2>&1
mkvmerge.exe -o "%%~na-2chremux.mkv" --no-audio "%%~na%%~xa" "%%~na.m4a" 1>>"%%~na.log" 2>&1
)
pause

[tebasuna51]

I preferred to mix the LFE with -3 dB when I used Azid.
Is the same than:
http://forum.doom9.org/showthread.ph...95#post1600695
If exist the parameter -mixlfe add:
FL' = ... + 0.7071 x LFE
FR' = ... + 0.7071 x LFE

[tebasuna51]

1) I prefer the Nero 7 AC3 decoder (if you have it installed), thus I force it over ffmpeg using the -nero switch. I have not had any problems with the "bug in disabling Nero DRC" in a few hundred encodes.
Apply DRC is not a problem when you want downmix, normalize and want more dialogs volume.
Is only a problem when you want preserve the original quality of your source.

2) I prefer -downStereo over -downDpl because it gives cleaner dialogues. Although -downDpl effects might sound a little better due to the 'matrix' echo effect in left channel, I found dialogues to be cleaner in -downStereo.
Of course -downStereo is preferred when you never play the audio with a 5.1 system with DPL decoder.

BTW:
- the echo effect must be in both left and right channels
- the contribution of Center channel (dialogues) is the same in -downStereo and -downDpl mix.

3) I prefer -mixlfe as it includes bass that gets picked up by the crossover on my 2.1 speaker set.
Maybe when you play with a 2.1 audio system, never recommended when play with TV speakers.
Add the LFE channel reduce the global volume of the rest of channels to avoid clip when normalize.
Also Dolby Digital say than mathematical add LFE to front channels can produce, sometimes, cancels or not desired effects.

4) I use -full to let eac3to pass qaac 64 bit-depth data since that is what is generated by the eac3to downmixer anyways. No need to do a down dithering operation since qaac will take care of that when it converts the audio data to floating point.
OK.
BTW preserve 64 bit precission is not very important after apply DRC, downmix and normalize, 24 bit integer is enough to convert to a lossy format.

5) I use qaac because it has --no-delay which removes the 44ms encoder delay introduced by all AAC encoders. Thus the audio is insured to be in sync with the original video.
NeroAacEnc put the delay in .m4a metadata and MkvMerge apply the needed delay in the AAC stream.

[Overdrive80]

Hi, I have .tsv file than if I rename to .ts, mediainfo show this details (http://pastebin.com/zJAGBM0d). But eac3to launch error. The video file is playback with mpc properly, premiere import without troubles. Mkvmerge import and all fine.

The source file is TV capture.

eac3to v3.27
command line: "C:\Program Files (x86)\MeGUI\tools\eac3to\eac3to.exe" "C:\Users\Isra\Desktop\INAZUMA ELEVEN_ '¡LA PRUEBA DEL CAPITAN!'(26022013_191942).TS" -progressnumbers -log="C:\Program Files (x86)\UsEac3to\UsEac3To.log"
------------------------------------------------------------------------------
The format of the source file could not be detected. <ERROR>

EDIT: DGindex is able to load too without troubles.

EDIT2: I attach example file (http://db.tt/H6OiU8p9)

EDIT3: I did contact with support department of device, and they said that I can handle tsv files with avidemux, and its true.

[PowerGamer]

When I want to extract DTS-HD MA 7.1 audio track from BluRay disc in its "original form" (for muxing into mkv later) I specify .dtsma extension for the output file (for ex.: "eac3to.exe 00800.mpls 3:audio.dtsma") and I don't need to have ArcSoft DTS Decoder installed, right? In other words, ArcSoft DTS Decoder needs to be installed only if I want eac3to to convert DTS-HD MA 7.1 audio track into some other format?

[tebasuna51]

@PowerGamer

Yes, is correct.

You can use .dts or .dtshd extension also, instead .dtsma, the stardard dts is extracted only if you put the parameter -core.

[Anakunda]

HI! I have finally managed to make working ArcSoft DTS decoder with eac3to but to my surprise the result is unusable.
Firstly I don't understand why the intermediate WAV by ArcSoft is 2-3x smaller than WAV made by ffmpeg, and secondly the result otherwise retains all 7.1 channels but the actors voice is missing at all and moreover after muxing it with video track the sound is (in KMP) simply ugly. What's wrong with ArcSoft? (I used DtsDec.dll and dtsdecoderdll.dll from latest TMT for this).

eac3to v3.24
command line: D:\media\eac3to\eac3to.exe Source.dtshd stdout.wav -normalize
------------------------------------------------------------------------------
DTS Master Audio, 7.1 (strange setup) channels, 24 bits, 48kHz
(core: DTS-ES, 5.1 channels, 24 bits, 1509kbps, 48kHz)
CAUTION: Decoding this track with ArcSoft results in low volume. <WARNING>
Decoding with ArcSoft DTS Decoder...
Applying RAW/PCM delay...
Writing WAV...
Creating file "stdout.pass1.wav"...
The original audio track has a constant bit depth of 24 bits.
Starting 2nd pass...
Reading WAV...
Reducing depth from 64 to 24 bits...
Writing WAV...
Applying 0.92dB gain...
Creating file "stdout.wav"...
The processed audio track has a constant bit depth of 24 bits.
eac3to processing took 50 minutes, 16 seconds.
Done.

[Boulder]

Have you tried updating eac3to for starters? You are using an old version, the latest one is v3.27.

[Anakunda]

No yet, I noticed there's new version just now. I'll made a rip with new e3t and will see..

[Anakunda]

Wov it works with 3.27 :rolleyes: now why the temporary wav file for 2hour 8 channel audio track only has about 8GiB when the same 6 channel temporary wav from same track had almost 20gigs with ffmpeg? Does Arscoft decode lossy? This is what mediainfo writes about it:
Audio
Format : PCM
Format settings, Endianness : Little
Format settings, Sign : Signed
Codec ID : 00001000-0000-0100-8000-00AA00389B71
Bit rate mode : Constant
Bit rate : 9 216 Kbps
Channel(s) : 8 channels
Channel positions : Front: L C R, Side: L R, Back: L R, LFE
Sampling rate : 48.0 KHz
Bit depth : 24 bits

And why the encoded track is so quiet? This in log:

eac3to v3.27
command line: D:\media\eac3to\eac3to.exe Source.dtshd stdout.wav -normalize
------------------------------------------------------------------------------
DTS Master Audio, 7.1 (strange setup) channels, 24 bits, 48kHz
(core: DTS-ES, 5.1 channels, 24 bits, 1509kbps, 48kHz)
CAUTION: Decoding this track with ArcSoft results in low volume. <WARNING>
Decoding with ArcSoft DTS Decoder...
Applying RAW/PCM delay...
Writing WAV...
Creating file "stdout.pass1.wav"...
The original audio track has a constant bit depth of 24 bits.

Can this be fixed please?

[rapscallion]

HI! I have finally managed to make working ArcSoft DTS decoder with eac3to but to my surprise the result is unusable.

You don't say which version of AS decoder you're using.

Those files, from Ver v1.1.0.0, work perfectly. Anything newer, not so much.

Plus, 20gb for just an audio track. That's not right, considering the complete file, with video, is 27gb !

Edit: is source.dtshd a result of encoding via DTS Master Audio Suite? If so, you have to remove the header hex by demuxing, via Tsmuxer , before running through eac3to.

[Anakunda]

You don't say which version of AS decoder you're using.

Those files from Ver v1.1.0.0, work perfectly. Anything newer, not so much.

Plus, 20gb for just an audio track. That's not right !

Yes I got such a sizes with stadard dts decoder, don't know where's the difference. Maybe it was floating point while AS does 24 bit fixedpoint? I don't remember the std decoder output params already.

And where can I get the old AS decoder files which "work perfectly"? I extracted the mines from Totalmediatheatre 6.0.1.119 installation. Are these ok?

Edit: is source.dtshd a result of encoding via DTS Master Audio Suite? If so, you have to remove the header hex by demuxing, via Tsmuxer , before running through eac3to.
No this was extracted right from bluray disc by eac3to. Do I need to remove the header anyway?

[rapscallion]

No this was extracted right from bluray disc by eac3to. Do I need to remove the header anyway?

No, however, I would rename it to "dts"

PM sent

[rapscallion]

I just realized you were decoding to wav/pcm, so 8gb would not be unusual.

What are you trying to accomplish ?

[Sparktank]

I extracted the mines from Totalmediatheatre 6.0.1.119 installation. Are these ok?

For ArcSoft 1.1.0.0, the "dtsdecoderdll.dll" has two different dates you should watch out for.

(From TMT 2.1.6.120)
Good: v1.1.0.0 25/04/2008
Bad: v1.1.0.0 21/04/2008

More info here:
http://forum.doom9.org/showthread.php?p=1266679#post1266679

[DarkSpace]

now why the temporary wav file for 2hour 8 channel audio track only has about 8GiB when the same 6 channel temporary wav from same track had almost 20gigs with ffmpeg?

I just did some rough calculations, but (8 GB / 24 bit) * 64 bit equals roughly 21 GB, so I'd guess that ffmpeg decodes the lossy DTS core track at 64 bit floating point precision whereas ArcSoft decodes the lossless DTS-HD track at its specified precision of 24 bit. Comparing file sizes is not always a reliable way to find out which file is better, it only serves to show which file holds more (theoretical) information.

[Anakunda]

I guessed that libavcodec uses floating point too, this may explain the huge filesize difference. But I still get awfully sounding conversion on output with
eac3to Source.dtshd stdout.wav -normalize | qaac --ignorelength -o Destination.m4a - whichever ArcSoft files I use (now testing ArcSoft package 1.1.0.0)
What I don't understand if I open the dtshd file in foobar2000 using eac3to as decoder wrapper, the file plays very good, no annoying hiss and not quiet even in 7.1 layout. Simply no artifacts. Why the conversion is so poor, I use the same command line???

[rapscallion]

As I asked above, just what is it that you're trying to accomplish????

@Sparktank, he has v1.1.0.0 25/04/2008

[Anakunda]

I want to convert dts hd 8 channels track to aac 8 channels track

[rapscallion]

So, I assume, that you can play the BD on an Apple device.

I use Arcsoft Media Coverter for that. One step process and works every time.

Any further with eac3to, maybe someone else can chime in.

[Anakunda]

Oh yes I dont have apple device :cool:.
Maybe someone else can explain why the 8channels aac sound so distorted.

[Boulder]

I have sometimes had problems with outputting stuff if ffdshow is set to output 32bit floating point audio, there's static and distortion. Have you checked that the problem is not in the decoder of the result instead of something else. Maybe you could put a sample somewhere so we can test it too.

[DarkSpace]

Maybe someone else can explain why the 8channels aac sound so distorted.
As mentioned in the comment by Boulder, you need to make sure that isn't just your decoder making things sound ugly. Aside from that, I propose that you separate your command line into two steps and use an intermediate WAV file and play that back to verify that it's not eac3to that's making your audio sound ugly (remove the -full switch if you don't have enough free space).
eac3to "Source.dtshd" "Intermediate.wav" -normalize -full
qaac --ignorelength -o "Destination.m4a" "Intermediate.wav"
Also, I don't know how qaac works, so make sure that it's getting the input it expects (e.g. don't give it floating point data when it expects integer).

[rapscallion]

Oh yes I dont have apple device :cool:.
Maybe someone else can explain why the 8channels aac sound so distorted.

OK, you've really got me curious, what's the point/purpose of doing this conversion?

[Chumbo]

I have a few dts files I'm converting to ac3 but the ArcSoft decoder is reporting an error. What are the best ways to get the file converted using an alternate method? For now, I'm basically checking if a failure occurs on the default setting, then use -sonic and if that fails, use -nero. Is this a good approach? Seems to be working fine. Am I using the recommended next-in-line to the ArcSoft decoder by going to sonic first and then nero? Thanks.

[Furiousflea]

I have a few dts files I'm converting to ac3 but the ArcSoft decoder is reporting an error. What are the best ways to get the file converted using an alternate method? For now, I'm basically checking if a failure occurs on the default setting, then use -sonic and if that fails, use -nero. Is this a good approach? Seems to be working fine. Am I using the recommended next-in-line to the ArcSoft decoder by going to sonic first and then nero? Thanks.

Yea, Arcsoft -> Sonic -> Nero

(Best avoid nero altogether ;) for any DTS-HD)

[Chumbo]

Yea, Arcsoft -> Sonic -> Nero

(Best avoid nero altogether ;) for any DTS-HD)
Thanks, that's what I thought. Thankfully, not a dts-hd track. ;)

[Anakunda]

I propose that you separate your command line into two steps and use an intermediate WAV file and play that back to verify that it's not eac3to that's making your audio sound ugly (remove the -full switch if you don't have enough free space).
I did so now and got strange results.
The dtshd track reported length by MI is none. In foobar2000 it's reported only 1h02m which is even not a half of true length.

ediainfo for dtshd: Audio
Format : DTS
Format/Info : Digital Theater Systems
Format profile : MA / Core
Mode : 16
Format settings, Endianness : Big
Bit rate mode : Variable
Bit rate : Unknown / 1 509 Kbps
Channel(s) : 8 channels / 6 channels
Channel positions : Front: L C R, Side: L R, Back: L R, LFE / Front: L C R, Side: L R, LFE
Sampling rate : 48.0 KHz
Bit depth : 24 bits
Compression mode : Lossless / Lossy

Now I decoded it to wav using eac3to source.dtshd intermediate.wav -normalize -full

and got this 22Gigs big WAV file. If I open it in foobar2000 it seems to play all channels properly but is reported length only 10m29s

Audio
Format : PCM
Format profile : Float
Codec ID : 00001000-0000-0300-8000-00AA00389B71
Codec ID/Hint : IEEE
Duration : 2h 6mn
Bit rate mode : Constant
Bit rate : 24.6 Mbps
Channel(s) : 8 channels
Channel positions : Front: L C R, Side: L R, Back: L R, LFE
Sampling rate : 48.0 KHz
Bit depth : 64 bits
Stream size : 21.8 GiB (100%)

Also trying to convert this to aac results in only 10m29s track :(

[DarkSpace]

The dtshd track reported length by MI is none. In foobar2000 it's reported only 1h02m which is even not a half of true length.
I have no idea exactly what foobar2000 is doing, but I have read several times already that guessing the duration of an Elementary Stream with variable bitrate can be tricky (there is no container to tell you the true duration, after all).

Now I decoded it to wav using eac3to source.dtshd intermediate.wav -normalize -full and got this 22Gigs big WAV file. If I open it in foobar2000 it seems to play all channels properly but is reported length only 10m29s
Some players have problems with WAV files larger than 2 GB. If your player doesn't play the whole file, try using the .w64 format for conversion instead of .wav (no worries, both formats are lossless). If your player plays the whole file but displays a wrong duration, you don't have to convert to .w64, although of course it shouldn't harm things.

Also trying to convert this to aac results in only 10m29s track :(
Same as above, WAV files greater than 2 GB may be problematic. However, before I mention that you may need to look for a switch to convert the whole file to AAC, make sure the --ignorelength parameter is present. Alternatively, you can also use the .w64 approach mentiond above, if qaac can handle .w64 files.

[dvbt]

hello friends when I convert a file into ac3 EAC3 here is the error in EAC3 tools


eac3to v3.27
command line: "C:\Users\le pc de jax\Desktop\e\eac3to.exe" "D:\PVR SMART DVB\smartdvb DVB-C-DVB-T\numero23.eac3" "D:\PVR SMART DVB\smartdvb DVB-C-DVB-T\15.ac3" -192
------------------------------------------------------------------------------
E-AC3 Surround, 2.0 channels, 0:42:52, 128kbps, 48kHz, dialnorm: -23dB
Removing E-AC3 dialog normalization...
Decoding with libav/ffmpeg...
Encoding AC3 <192kbps> with libAften...
[libav] exponent out-of-range <WARNING>
[libav] error decoding the audio block <WARNING>
[libav] exponent out-of-range <WARNING>
[libav] error decoding the audio block <WARNING>
Creating file "D:\PVR SMART DVB\smartdvb DVB-C-DVB-T\15.ac3"...
[libav] exponent out-of-range <WARNING>
[libav] error decoding the audio block <WARNING>
[libav] exponent out-of-range <WARNING>
[libav] error decoding the audio block <WARNING>
[libav] exponent out-of-range <WARNING>
[libav] error decoding the audio block <WARNING>
eac3to processing took 20 seconds.
Done.

[tebasuna51]

Where is the <ERROR>?
I only see <WARNING>'s

The proccess can finish OK, maybe with some gaps because imperfect TV capture.
The output AC3 have similar length and play OK?

[dvbt]

hello the ts and perfect without any error

[dvbt]

here is the log of ts
# # File = D: \ SMART PVR DVB \ POUCHIN TV DVB-T PVR \ NUMBER 23 2013-01-14 12-06-13.ts
PIDs included (* = PCR) = 210, 220 *, 230, 231, 232, 240, 241
PIDs found root = 0
First and last PCR = 24:16:01.8, 25:53:44.3 (in pid 220)
Duration = 01:37:42.5
-------------------------------------------------- -------------------------------------------------
Current audit
Made.
Summary for this file:
Number of packets = 23201252
Continuity errors = 0
Error flags = 0
Error packets = 0
Total time = 01:37:42.5
Note that this term is only based on the time marker
found in the TS stream. If parts are missing in the file
output, the displayed value will likely be inaccurate.
verification is complete



I have a glitch when the bed by my EAC3 vlc here is the the problem is the first to second http://dl.free.fr/getfile.pl?file=/CeIa83pB

[tebasuna51]

Yes, there are a audible glitch at the begining. That hapens with TV capture.
Decode to wav, edit the problem and recode after to ac3.

[Boulder]

Does Sonic downmix 7.1 material to 5.1 when decoding or what does the "decodes only 5.1" mean in the first post? I have a 7.1 DTS-HD MA track which refuses to be decoded without errors with Arcsoft and I'd like to use the whole lossless track for encoding to AAC.

[tebasuna51]

I don't use Sonic decoder but you can try decoding this Channel_Test_DTS-MA_7.1 (http://www.sendspace.com/file/fva4xn) (only 1.3 MB) and see what hapens.
All channels have a max peak at 100% (LFE only 50%), analyzing the output wav you can know how work the decoder.

[Infineon]

I've been trying to convert DTS-HD to lossless WAV/FLAC for my media library. I've done some reading up on the requirements and have been using eac3to with the blu-ray stream extractor GUI but cannot get a suitable DTS-HD decoder working and just end up with 48Khz output from the DTS core. I gather you need to source files from an old version of TMT (2?) for the setup to work. I've been unable to find this, even a trial download.

Can anyone point me / msg me where I might get the required DLLs or is there another way to do this now?

[LigH]

Looking for the ArcSoft Codec Unlocker / ASAudioHD+ (http://forum.doom9.org/showthread.php?p=1392750)?

Or simply install RipBot264 (http://forum.doom9.org/showthread.php?t=151692)... :rolleyes:

[Furiousflea]

I use Arcsoft decoder to decode DTS-MA and encode to FLAC using eac3to.

However, I just read this thread and the person talks about changing default settings for the Arcsoft decoder...
http://forum.doom9.org/showthread.php?p=1392750

This makes me worry that the FLAC that is output from decoding the DTS-MA track has dynamic range compression applied.

Could someone confirm that this is or isn't the case and there is no need to mess around with any settings to use the Arcsoft decoder correctly (apart from having it recognized by eac3to of course)....

:eek:

Thanks in advance.

[nevcairiel]

eac3to should set the options for the decoder, you don't need to worry about it.

[Furiousflea]

eac3to should set the options for the decoder, you don't need to worry about it.

Thanks mate...was worrying about hundreds of hours of work ahead :o

[Infineon]

I got a trial copy of TMT 3 Platinum Retail 3.0.1.160 and installed it. Got RipBot264 setup and I copied the required ax and DLL files into the eac3to folder from the RipBot264 Tools\ax\ArcSoft DTS Decoder\ folder. I then registered the ASAudioHD.ax file within the eac3to folder (confirmed registered successfully) but still I get:

"[a02] The ArcSoft and Sonic decoders don't seem to work, will use libav instead."

The version of the dtsdecoderdll.dll currently in my eac3to folder (coped from RipBot) is 1.1.0.0. The CheckActivate I copied over is 1.0.0.2.

Any ideas where I've screwed up?

[filler56789]

^ @Infineon: make sure your eac3to folder is included in the PATH (http://en.wikipedia.org/wiki/PATH_(variable)) (environment-variable).

[Infineon]

Thanks I've just added it but it hasn't made a difference. This is frustrating :(

[heerschop]

Thanks I've just added it but it hasn't made a difference. This is frustrating :(

In order, for me, to get arcsoft working with EAC3to I also have additional files in my arcsoft directory. I checked and when I remove one of these files, the decoder doesn't work anymore.
Here is the list of all the arcsoft decoder files:
- ASAudioHD.ax
- checkactivate.dll
- DtsDec.dll
- dtsdecoderdll.dll
- MagCore.dll
- MagPCMac.dll
- MagUIEngine.dll
- MagUIInter.dll

regsvr32.exe C:\xxx\Arcsoft\ASAudioHD.ax
add to path => C:\xxx\Arcsoft and Reboot

Hope this will help you,

Greetz

[Infineon]

In my setup I have copied the files from a trial version of ArcSoft into my eac3to.exe folder root (rather than into an 'ArcSoft' sub-folder for example). The only file I didn't copy from your list is 'DtsDec.dll'. That wasn't mentioned in any of the setup guides I've seen.

I think my biggest worry is that I don't have the correct versions of these files as I wasn't able to source the earlier trial downloads of TMT (versions 2 or 3). Would someone be kind enough to zip a copy of these and provide? (So I can be sure this isn't the issue). The version of eac3to I'm using is: 3.27.0.0 but I'm running it using version 3 of the 'HdBrStreamExtractor'.

[Overdrive80]

@Infineon @heerschop Please, see rule 6.

You have that put .dlls files in system32 or syswow64. If you OS is 32 bit then system32, and if is 64 bit in both, syswow64 and system32

[Nico8583]

Hi :)
Could you implement MKV AVC/MVC on next release ?
Thanks !

[LigH]

How exactly is AVC/MVC related to audio?