eac3to - audio conversion tool

[iSeries]

tebasuna51, thanks! The resulting .wav files seem to be 32 bit float though (from BeHappy), what do I need to add to the scripts to get 24 bit?

Thanks again!

[b66pak]

add:

ConvertAudioTo24bit()
_

[iSeries]

Thanks guys!

...is it ok to reduce the bit depth of the new .wav's from 32 bit float back to 24 bit with eac3to? or is it more recommended to do this with BeHappy?

[iSeries]

Also, do I need to normailize ALL the wav's that will make up the 5.1, or just the two created with BeHappy?

[tebasuna51]

Well, nothing is perfect. You can't obtain always the same global audio volume with 7 speakers than with 6.

If you want preserve the balance volume betwen front and back channels maybe you need make this mix:

BL = SL + 0.7071 x BC
BR = SR + 0.7071 x BC

preserve the 32 bit float and check if there are values greater than 0 dB, if yes you need use the normalize only over BL-BR and lose some back volume, if not you don't need the normalize and the balance volume is ok.

[iSeries]

Thanks, like this?

sl = WavSource("D:\...\SideLeft.wav")
bc = WavSource("D:\...\BackCenter.wav")
MixAudio(sl, bc, 0.5858, 0.7071)

[kypec]

Thanks, like this?

sl = WavSource("D:\...\SideLeft.wav")
bc = WavSource("D:\...\BackCenter.wav")
MixAudio(sl, bc, 0.5858, 0.7071)
Almost correct but the SL / SR channels must be factored by 1, not 0.5858.
Replace your last line with this:MixAudio(sl, bc, 1.0, 0.7071)

[iSeries]

Thanks! Sorry one last question lol, how can I check if clipping has occured? Am I right in thinking that if I open the wav in Audacity and nothing goes beyond 1.0 or -1.0 then there is no clipping?

Also, I don't really understand any of this so a brief explanation on the difference between

MixAudio(sl, bc, 0.5858, 0.4142)

and

MixAudio(sl, bc, 1.0, 0.7071)

...would be greatly appreciated :thanks:

[tebasuna51]

..., how can I check if clipping has occured? Am I right in thinking that if I open the wav in Audacity and nothing goes beyond 1.0 or -1.0 then there is no clipping?

You can use Audacity -> View -> Show clipping

MixAudio(sl, bc, 0.5858, 0.4142)
make:
BL = 0.5858 x SL + 0.4142 x BC
the audio never can be greater than 1.0 (0 dB), without clip

and

MixAudio(sl, bc, 1.0, 0.7071)
make:
BL = 1.0 x SL + 0.7071 x BC
the audio can be greater than 1.0, then you can have clipping.

[iSeries]

Many thanks. I used MixAudio(sl, bc, 1.0, 0.7071) for LOTR Fellowship 1/2, and this produced no clipping. The Two Towers 1/2 and Return of the King 1/2 did produce clipping using that so for these I re-did with MixAudio(sl, bc, 0.5858, 0.4142). I'm guessing thats all ok (Audacity shows no clipping for all wavs) :)

[Vasch the stampede]

eac3to v3.24
command line: eac3to "...\00010.m2ts" 1:"...\1.txt" 2:"...\2.h264" 4:"...\4.wav"
------------------------------------------------------------------------------
M2TS, 1 video track, 6 audio tracks, 18 subtitle tracks, 2:02:34, 24p /1.001
1: Chapters, 21 chapters
2: h264/AVC, 1080p24 /1.001 (16:9)
3: DTS Master Audio, English, 6.1 channels, 24 bits, 48kHz
(core: DTS-ES, 6.1 channels, 24 bits, 1509kbps, 48kHz)
4: DTS Master Audio, Italian, 6.1 channels, 16 bits, 48kHz
(core: DTS-ES, 6.1 channels, 16 bits, 1509kbps, 48kHz)
5: AC3 Surround, English, 2.0 channels, 192kbps, 48kHz, dialnorm: -27dB
6: AC3 Surround, English, 2.0 channels, 192kbps, 48kHz, dialnorm: -27dB
7: AC3 Surround, English, 2.0 channels, 192kbps, 48kHz, dialnorm: -27dB
8: AC3 Surround, English, 2.0 channels, 192kbps, 48kHz, dialnorm: -27dB
9: Subtitle (PGS), English
10: Subtitle (PGS), English
11: Subtitle (PGS), English
12: Subtitle (PGS), English
13: Subtitle (PGS), English
14: Subtitle (PGS), English
15: Subtitle (PGS), Italian
16: Subtitle (PGS), Italian
17: Subtitle (PGS), Italian
18: Subtitle (PGS), Italian
19: Subtitle (PGS), Italian
20: Subtitle (PGS), Italian
21: Subtitle (PGS), Danish
22: Subtitle (PGS), Finnish
23: Subtitle (PGS), Modern Greek
24: Subtitle (PGS), Icelandic
25: Subtitle (PGS), Norwegian
26: Subtitle (PGS), Swedish
Creating file "..\1.txt"...
[v02] Extracting video track number 2...
[a04] Extracting audio track number 4...
[a04] Remapping channels...
[a04] Decoding with ArcSoft DTS Decoder...
[a04] Writing WAV...
[v02] Creating file "...\2.h264"...
[a04] Creating file "...\4.wav"...
[a04] The original audio track has a constant bit depth of 16 bits.
[a04] Caution: The WAV file is bigger than 4GB. <WARNING>
[a04] Some WAV readers might not be able to handle this file correctly. <WARNING>
Video track 2 contains 176328 frames.
eac3to processing took 56 minutes, 13 seconds.
Done.

Why remapping channels?

[tebasuna51]

Because the DTS internal channel order is different than standard WAV order.

[der_andi]

Hi,
I noticed a performance decrease in reading the blu-ray structure compared to an older version of eac3to. Consider these automated commands:

eac3to.exe f:
eac3to.exe f: 1)
eac3to.exe f: 2)
...
eac3to.exe f: 5)

The output of the first line is used by a program in order to determine the number of titles on that disc. Let it be 5, for example. Then it analyses all titles automatically. Version 3.18 can do these commands in around 400 seconds, while the latest version 3.24 needs around 490. I get similar results with all my blu-rays.

Any ideas what's causing this additional time?


Best regards,
Andi

[NanoBot]

Hi,

not directly an eac3to question, but the reason for this question is the eac3to bug when downmixing 6.1 to 5.1 and the need to decode and downmix manually: I think and hope that you can help me with this question:

I found these channel orders for 5.1 respectively 6.1 flac files:

5.1 FL, FR, C, LFE, SL, SR
6.1 FL, FR, C, LFE, BC, SL, SR

Are these channels orders, which I got from ffdshow as audio audio processor from and searching for them in the internet, correct ?

Thx in advance, NanoBot

[tebasuna51]

But is possible also:

6.1 FL, FR, C, LFE, BL, BR, BC

but madshi know the problem, is only a bug that can be solved with a new version ...

[tooftah]

Hello,

can anyone tell me what decoder to use for 2-channel DTS Express tracks? I've tried using Arcsoft 1.1.0.0, 1.1.0.8 and Nero 7 decoders, and they all produce different results. Stereo WAVs from Arcsoft seem to be identical, except for 1 byte difference every few kilobytes (checked using HxD). Nero 7 gives a totally different stream. Which one to choose?

Can the difference be caused by different versions of Windows? I have Arcsoft 1.1.0.8 on a Windows 7 x86_64 PC and Nero 7 + Arcsoft 1.1.0.0 on a Windows XP x86 PC. However they both decode e.g. DTS-HD 5.1 files identically.

[Thunderbolt8]

afaik dts express is not supported by any of encoder (which eac3to uses)

edit: or I am wrong (guess I am) and its only that mkvmerge doensnt support muxing it.

[tooftah]

Thunderbolt8,

I was talking about DEcoders, that work with eac3to. I have a Stereo DTS Express track, which I want to decode using eac3to. And my problem is, that all three decoders output different WAVs. However, you are in a way right, because I intend to compress it and then mux to a matroska file.

Somewhere in the middle of this giant thread madshi talked about using Nero decoder on files, which Arcsoft doesn't handle. But all three decoders produce audio without an error, at least the program doesn't show any + I hear sound on playback, not silence or some random noise, etc.

[iSeries]

Hi,

Having a bit of difficulty downconverting the sound track to 'Faster' to 16-bit. After processing with the -down16 switch, the eac3to log indicates that the DTS-MA track is a constant 20-bit. Is there a way to get eac3to to downconvert 20-bit to 16-bit and not assume 24-bit?

[nurbs]

Hard to say without seeing the log, but it probably is a 24bit file, it's just that the actual audio data contained in it is only 20bit. I don't think the initial bitdepth has much influence on the speed of the processing, but tebasuna51 will probably know better than me. If your target format isn't lossless you can skip the downconversion to save time.

[iSeries]

You are right in that it is a 24-bit file and that the actual audio data is 20-bit. I'm asking if there is a way to down-convert to 16-bit from the 20-bit data. As it is, the -down16 switch treats the file as though it has 24-bits of audio data.

[tebasuna51]

I can't understand your problem, use -down16.

The uncompressed audio data are stored in 2 bytes for 16 bits and in 3 bytes for 24 bits, don't exist a especific 20 bits storage, when eac3to say is 20 bits means than the 4 less significants bits, from the 3 bytes/24 bits, are always 0.
Here down-convert from 24 or 20 bits is the same, forget the less significant byte.

[iSeries]

So this doesn't apply to 20-bit audio files padded to 24-bit?

http://forum.doom9.org/showthread.php?p=1404212&highlight=iseries#post1404212

http://forum.doom9.org/showthread.php?p=1404320&highlight=iseries#post1404320

[tebasuna51]

Sorry but I can't understand your question.

[nibus]

can you demux to a destination directory instead of the eac3to directory?

[der_andi]

can you demux to a destination directory instead of the eac3to directory?

Yes, you can. Here's an example for demuxing a video stream from a blu-ray:
eac3to.exe F: 1) 2: C:\my\destination\directory\movie.mkv

That's the simplest way. But if you demux several streams at once, you have to type the path for every stream. That's a lot of work.
Instead, you could append the directory of eac3to.exe to the PATH environment variable. Then eac3to works from every directory, e.g. your desired destination directory. And you can specify filenames as you've been doing all along.

[nibus]

Instead, you could append the directory of eac3to.exe to the PATH environment variable. Then eac3to works from every directory, e.g. your desired destination directory. And you can specify filenames as you've been doing all along.

Cool, I hadn't thought of this. thanks!

[TinTime]

Yes, you can. Here's an example for demuxing a video stream from a blu-ray:
eac3to.exe F: 1) 2: C:\my\destination\directory\movie.mkv


If you use something like this then eac3to will demux to the specified directory:

eac3to.exe source "C:\my\destination\directory\movie.*"

[Weirdo]

can you demux to a destination directory instead of the eac3to directory?I'd suggest Another EAC3to GUI Plus for that. Works beautifully.

[nibus]

If you use something like this then eac3to will demux to the specified directory:

eac3to.exe source "C:\my\destination\directory\movie.*"

Oh nice, that works even better. :)

[ramicio]

Any time I get movies with 6.1 DTS-HD tracks the resulting file is much longer than what it's supposed to be, and the sound is bad/wrong. This is with the Arcsoft decoder.

[nibus]

Any time I get movies with 6.1 DTS-HD tracks the resulting file is much longer than what it's supposed to be, and the sound is bad/wrong. This is with the Arcsoft decoder.

You need dtsdecoder.dll version 1.1.0.0 to properly decode 6.1. However 1.1.0.0 does NOT work correctly with 7.1 (* on non-standard channel mapping only), for that you need version 1.1.0.8. When you switch versions you will need to re-register ASAudioHD.ax by entering this into the command prompt:

regsvr32.exe ASAudioHD.ax

Both versions work with regular 5.1, and to my knowledge there is no need for any other versions.

[tebasuna51]

... However 1.1.0.0 does NOT work correctly with 7.1, for that you need version 1.1.0.8.
Please can you explain for what don't work 1.1.0.0 with 7.1, works fine for me.

[nibus]

Please can you explain for what don't work 1.1.0.0 with 7.1, works fine for me.

I think I read earlier in this thread that the channel mapping was off... I'll see if I can dig the specific post up.

edit: Here it is-

http://forum.doom9.org/showthread.php?p=1469584#post1469584

Yes, we need switch in this situation.
In case of non-standart scheme decoder 1.1.0.0 give garbage in some channels. So, i tried other two versions and they are decoded properly. If you have DTS-HD M.A.S. you can try yourself. If not, try decode test sample (http://www.mediafire.com/?kagbxe5zg7tdzg0) of DTS-HD 7.1 (scheme 2) and compare results.
AFAIK, there are only two differences between 1.1.0.0 and 1.1.0.8:
- 1.1.0.0 can decode DTS(-HD) 6.1/6.0 but can't decode non-standart 7.1
- 1.1.0.8 can't decode DTS(-HD) 6.1/6.0 but can decode non-standart 7.1
Both decode DTS(-HD) 1.0 correctly, unlike 1.1.0.7.

Only for non-standard channel mapping.

[ramicio]

I know we're not supposed to ask where to get these filters, but I'm going to just ask rhetorically...where are we supposed to find all these random older versions of this filter? Is there one version that works perfectly with every possible layout? Someone really needs to get on the ball and get DTS-HD decoding into libavcodec.

[tebasuna51]

1.1.0.0 work also with non-standard channel mapping without garbage.
Like don't exist channels Ls-Rs in WAV make a remux to BL-BR correct to my opinion.

[ramicio]

It's almost getting to be more worth the time to just fake-header a DTS-HD file and let the DTS suite decode the files. I really find that having to switch filter versions to be inexcusable (on Arcsoft's part, not madshi, he makes awesome stuff). I have no idea how I would make a 6.1 wav, though. There is no documentation on channel masks for wavavimux.

[Midzuki]

It's almost getting to be more worth the time to just fake-header a DTS-HD file and let the DTS suite decode the files. I really find that having to switch filter versions to be inexcusable (on Arcsoft's part, not madshi, he makes awesome stuff).

I agree. However I do not have the entire recipe for adding a new header (AND a new footer) to a "naked" :) DTS-HD file — would you mind sharing yours? :confused:

I have no idea how I would make a 6.1 wav, though. There is no documentation on channel masks for wavavimux.

Channel Mask for "normal" 6.1 audio = 319.

CHANNEL NAME --- Decimal Value
¯¯¯¯¯¯¯¯¯¯¯¯ ¯¯¯¯¯¯¯¯¯¯¯¯¯

FRONT_LEFT 1
FRONT_RIGHT 2
FRONT_CENTER 4
LOW_FREQUENCY 8
BACK_LEFT 16
BACK_RIGHT 32

FRONT_LEFT_OF_CENTER 64
FRONT_RIGHT_OF_CENTER 128

BACK_CENTER 256
SIDE_LEFT 512
SIDE_RIGHT 1024

*************************************

TOP_CENTER 2048
TOP_FRONT_LEFT 4096
TOP_FRONT_CENTER 8192
TOP_FRONT_RIGHT 16384
TOP_BACK_LEFT 32768
TOP_BACK_CENTER 65536
TOP_BACK_RIGHT 131072
RESERVED 262144

[ramicio]

I make a mono wav that contains silence, make it substantially longer than the track we're trying to work with (we'll call this "the useless source") I need to decode, encode a track with the exact specs as the useless source, copy the first 140 bytes in a hex editor, not the last character and what its pattern of character looks like before that character, in the useless source, cut up to the last character of our nice header, and copy INTO the beginning of the useless source. It's very time consuming, but since it's a DTS tool, and all things most likely come from the tool, there can't be anything it can't handle.

Now I see the scheme for the channel mask, it's just sums! I never saw the chart you posted though, so thank you!

[ramicio]

Would there ever be any chance of taking multiple mono WAV files in to create a single multi-channel file? I can't find any decent and sure-fire way to do this with any other utility. WavAVImux to graphedit with wavdest, you are limited to a 4 GB output. Avisynth and VirtualDub, save WAV, can break the 4 GB barrier, but what about weird channel mappings such as 6.1? The there's wavewizard which I can't even begin to figure out, and have no idea about size limitations or channel mapping.

[Midzuki]

^ wavi.exe:


WAVI v1.06F - (c) 2k7 Tamas Kurucsai, with lots of help from tebasuna51
Licensed under the terms of the GNU General Public License.

This utility extracts the first uncompressed PCM audio track
from an AVI file and saves it to a WAV file. This is not quite
useful for most AVI files since they usually contain some kind
of compressed audio, but it can come handy when it's needed
to save an audio track from AviSynth.

Usage: WAVI <avi-file> [ <wav-file> [ /R | /X | /M <mask> ] ]

If <avi-file> is a valid AVI file which contains PCM audio,
the audio track will be written to <wav-file> as a WAV file.
If '-' is passed as <wav-file>, the WAV file will be written
to the standard output.
If <wav-file> is not given, only information will be printed
about the audio track.

On success, the exit code will be 0 and the first line printed
to the standard error will look like the following:

Found PCM audio: <c> channels, <r> Hz, <b> bits, <l> seconds.

where <c> is the number of audio channels, <r> is the sampling
rate, <b> is the number of bits per sample and <l> is the length
of the track in seconds.

If the audio track contains floating-point samples, the next line
printed to the standard error will be:
Audio track contains floating-point samples.

If an error occurs, the exit code will be 1 and some useful
error message will be printed to the standard error.

WAV files larger than 4 GB may be created. However,
such WAV files are non-standard and
may not be handled correctly by some players and encoders.
A warning will be printed to the standard error when such a WAV file is created.

WAVI accepts the following options:

/R - Write a raw file of samples without the WAV header.

/X - Write an extended WAV header containing the default
channel mask for multi-channel audio.

/M <mask> - Write an extended WAV header containing the
specified channel mask for multi-channel audio.

The default channel masks are:

Mask Chan. MS channels Description
---- ----- ------------------------- ----------------
4 1 FC Mono
3 2 FL FR Stereo
259 3 FL FR BC First Surround
51 4 FL FR BL BR Quadro
55 5 FL FR FC BL BR like Dpl II (without LFE)
63 6 FL FR FC LF BL BR Standard Surround
319 7 FL FR FC LF BL BR BC With back center
255 8 FL FR FC LF BL BR FLC FRC With front center left/right

Some other common channel masks:

Mask Chan. MS channels Description
---- ----- ------------------------- ----------------
7 3 FL FR FC
263 4 FL FR FC BC like Dpl I
271 5 FL FR FC BC LF
59 5 FL FR BL BR LF



but what about weird channel mappings such as 6.1?

Why do you think 6.1 is "weird"? :confused: IMHO, the only "weird" audio channels are the infamous/Sonyc Front-Left-Of-Center and Front-Right-Of-Center.

The there's wavewizard which I can't even begin to figure out, and have no idea about size limitations or channel mapping.

Me too hates wavewizard, you are not alone :D

[tebasuna51]

WaveWizard works fine for me but don't fill the MaskChannel field (always 0) then you need some other util like:

wavfix
Usage:
WavFix <input.wav> [output.wav] [-ignorelength] [-o #] [-m #] [-i/f #] [-c #] [-s #]

Where:
<input.wav> wav or w64 up to 8 channels, int or float, 8-16-24-32-64 bits,
PCM/WAVE_FORMAT_EXTENSIBLE/W64, any extra chunks, allowed > 4GB.
To use STDIN use - as input filename.

[output.wav] If not present, input.wav is used suffixed with: _fix
The header can be selected, see -o parameter.

[-ignorelength] If present the length in wav header is ignored, useful for wav
> 2/4 GB. Problem: extrachunks at end of file treated as data

[-o #] Output header. 0=RAW, 1=WAV(Default), 2=RF64, 3=W64

[-m #] MaskChannel. Force output WAVE_FORMAT_EXTENSIBLE. 0 to default

If any of the next parameters are present the input is considered RAW:
[-i #] BitsPerSample Integer. Default 16. Valid 8, 16, 24, 32

[-f #] BitsPerSample Float. Valid 32 or 64.

[-c #] NumChannels. Default 2. Valid only 1 to 8.

[-s #] SampleRate. Default 48000 Hz. Any value is allowed.

[EDIT] wavfix.exe included in wav2Util.7z here (https://forum.doom9.org/showthread.php?p=1624209#post1624209)

[ramicio]

Wavi doesn't work for me once an AVI file is created with wavavimux, even for ones that are under 2 GB. It used to, now it doesn't. It simply says there is audio in there, but it's not PCM. I don't understand how to use wavewizard, and I don't understand what that wavfix thing would even do. I just don't get how more people, or anyone at all has asked about eac3to being able to accept multiple mono wavs as input.

[Midzuki]

Wavi doesn't work for me once an AVI file is created with wavavimux, even for ones that are under 2 GB. It used to, now it doesn't. It simply says there is audio in there, but it's not PCM.

Thanks for useful the info --- I wasn't aware of that, because, until now, I had used wavi.exe only for .AVS files. Anyway, here is the workaround:

<intermed.avs>
DirectShowSource("audio-only.avi")
</intermed.avs>

wavi intermed.avs final.wav -M ###

Still, I don't know what you intend to do by creating a real multichannel .WAV greater than 4GB :confused: Applications like NeroAacEnc, WavPack, Aften, whatever, will happily encode from the stdin, and commercial compressors, by default, work with mono .WAVs :confused:

[ramicio]

Because, as previously stated, I have problems with certain channel schemes of DTS-HD Master, and I'm not going to try to hunt down the internet for specific versions of filters that I'll never find. So I must use the DTS suite to decode these weird ones into mono wavs. My end goal is a 5.1 FLAC. So I need to feed eac3to a multichannel WAV to downmix into 5.1 channels.

[Midzuki]

Ah, I see, at last :o

Hmmm, IMHO there really is no good reason why eac3to doesn't support .AVI as input. :confused:

It's a shame that neither sox, nor wavi, can output .W64 or .RF64 :(

[ramicio]

That would be a nice step forward if it supported AVI as an input. But a lot of things still have trouble with opening an AVI with only audio.

[tebasuna51]

... So I must use the DTS suite to decode these weird ones into mono wavs. My end goal is a 5.1 FLAC. So I need to feed eac3to a multichannel WAV to downmix into 5.1 channels.

eac3to can downmix the 7.1 L,R,C,LFE,Lss,Rss,Lsr,Rsr DTS channels to 5.1, but not 7.1 L,R,C,LFE,Ls,Rs,Lsr,Rsr.

How want you downmix the Ls,Rs,Lsr,Rsr to only 2 Surround channels? Please put the coefficients X1, X2 in:

SL = X1 x Ls + X2 x Lss
SR = X1 x Rs + X2 x Rss

I can make an avs file with your mix and can be used by BeHappy (or wavi) to obtain the 5.1 flac

[ramicio]

I'm fine with the way eac3to handles 7.1 stuff. Never had a problem. It's these pesky 6.1 I come across now and then. I like eac3to because it detects clipping. Would it be possible for avisynth to turn out 6.1?

[hello_hello]

Does ffdshow decode 6.1 DTS-HD? If so you could open the DTS file via DirectShow, use ffdshow's mixer matrix to mix it to stereo, 5.1ch, or any way you like. From there, any encoder which will encode via DirectShow should be able to do the job.

I sometimes use ffdshow to mix to stereo (for those times when the audio track starts off stereo then switches to 5.1ch, and foobar's internal 5.1ch mixer has a hissy fit), while using foobar2000 to encode to MP3. Assuming ffdshow decodes 6.1ch DTS, I can't think of a reason why you couldn't use ffdshow to mix it to 5.1, then use foobar2000 to encode the output directly to flac.

Or, if you don't want to use ffdshow's mixer, you could leave it disabled and get foobar200 to spit out a multichannel wave file.

You'd need the foobar DirectShow decoder plugin.