eac3to - audio conversion tool

[jlw_4049]

We've recently discovered removing dialNorm from a truehd track during demux actually edits the header in the way the Atmos data is no longer valid. It's still there, but it cannot be used by any official tools or players.

Is there any way to restore dialNorm to normal?

I confirmed this by demuxing with eac3to with -keepDialnorm and without it and the only one that actually worked with the Atmos channels was the -keepDialnorm. I'm not sure which header is being edited in the file itself, but editing it with a hex editor to match the one with -keepDialnorm allows the Atmos to work correctly again. However, I can't do this across other files as easily.

Is there a way to do this with eac3to or any other tool?

[tebasuna51]

@jlw_4049
Yes, to preserve the Atmos data you must extract it with -keepDialnorm.

And not, I don't know a tool to restore the dialNorm.

[jlw_4049]

@jlw_4049
Yes, to preserve the Atmos data you must extract it with -keepDialnorm.

And not, I don't know a tool to restore the dialNorm.

Do you think it's possible for such a tool to be designed? As the data is still in the file, just seems like the header of the file has been modified. Not allowing the atmos to be decoded.

[Nikos]

With the following line:
eac3to audio.thd audio.wav
eac3to removes the Dialogue Normalization metadata.

The question is what happens to Dynamic Range Compression metadata?

[tebasuna51]

The question is what happens to Dynamic Range Compression metadata?

The old libav decoder in eac3to output the same volume than updated ffmpeg, even when I try use -drc_scale 0 with the message:

Codec AVOption drc_scale (percentage of dynamic range compression to apply) specified for input file #0 (D:\Test\chan-check-5.1.thd) has not been used for any stream. The most likely reason is either wrong type (e.g. a video option with no video streams) or that it is a private option of some decoder which was not actually used for any stream.

I try with many thd samples with the same message, then I don't have a sample with DRC, or ffmpeg ignore that metadata if exist.
I can't check it.

[Nikos]

After testing I came to the conclusion:
With a .thd sound, the ffmpeg ignores the DN and the DRC.

Center channel .thd
drc_scale 0.0 --> RMS -28.9576 dB
drc_scale 1.0 --> RMS -28.9576 dB

With a .ac3 sound (core from .thd), the ffmpeg ignores the DN but applies the DRC.

Center channel .ac3 core
drc_scale 0.0 --> RMS -29.0598 dB
drc_scale 0.5 --> RMS -30.2756 dB
drc_scale 1.0 --> RMS -31.1711 dB

[tebasuna51]

The volume level of ac3 drc_scale 0.0 (-29 dB) is the same than the thd, then seems the decode is correct without atenuation.

[nevcairiel]

Dialog Normalization in AC3 is only used if you set target_level to adjust the target volume of the decoded stream. Its how the AC3 spec wants to control a unified volume, by including metadata to let you request a reference level.

[Nikos]

The ffmpeg decoding is correct when with drc_scale 0 we have no compression and with drc_scale 1 we have full compression :)

For downmix 5ch --> 2ch i prefer audio with DRC=1.

I don't know what it is ac3 target_level...

[Balling]

The old libav decoder in eac3to output the same volume than updated ffmpeg, even when I try use -drc_scale 0 with the message:



I try with many thd samples with the same message, then I don't have a sample with DRC, or ffmpeg ignore that metadata if exist.
I can't check it.

FFmpeg does NOT support TrueHD DRC, it just outputs lossless stream, that is the same (aside from the lossy) as EAC3 with -drc_scale 0. Also, FFmpeg does not support it anywhere with encoding.

But EAE and microsoft and others support decoding TrueHD (now lossily) with DRC.

[Balling]

After testing I came to the conclusion:
With a .thd sound, the ffmpeg ignores the DN and the DRC.

Center channel .thd
drc_scale 0.0 --> RMS -28.9576 dB
drc_scale 1.0 --> RMS -28.9576 dB

With a .ac3 sound (core from .thd), the ffmpeg ignores the DN but applies the DRC.

Center channel .ac3 core
drc_scale 0.0 --> RMS -29.0598 dB
drc_scale 0.5 --> RMS -30.2756 dB
drc_scale 1.0 --> RMS -31.1711 dB

drc_scale is not EVEN an option for truehd decoder!!
Compare commands ffmpeg -h decoder=truehd and ffmpeg -h decoder=eac3

[tebasuna51]

FFmpeg does NOT support TrueHD DRC, it just outputs lossless stream, that is the same (aside from the lossy) as EAC3 with -drc_scale 0. Also, FFmpeg does not support it anywhere with encoding.
...
drc_scale is not EVEN an option for truehd decoder!!
Compare commands ffmpeg -h decoder=truehd and ffmpeg -h decoder=eac3
Thanks to confirm the ffmpeg behaviour. Not always is easy know how ffmpeg work.

Decoder eac3 [ATSC A/52B (AC-3, E-AC-3)]:
(E-)AC3 decoder AVOptions:
-cons_noisegen <boolean> .D..A...... enable consistent noise generation (default false)
-drc_scale <float> .D..A...... percentage of dynamic range compression to apply (from 0 to 6) (default 1)
-heavy_compr <boolean> .D..A...... enable heavy dynamic range compression (default false)
-target_level <int> .D..A...... target level in -dBFS (0 not applied) (from -31 to 0) (default 0)
-downmix <channel_layout> .D..A...... Request a specific channel layout from the decoder

The default -drc_scale 1 is correct for a player, but for a decoder the default must be 0.

But EAE and microsoft and others support decoding TrueHD (now lossily) with DRC.

Bad news if there aren't a option to override the DRC.

[resolution]

We've recently discovered removing dialNorm from a truehd track during demux actually edits the header in the way the Atmos data is no longer valid. It's still there, but it cannot be used by any official tools or players.

Is there any way to restore dialNorm to normal?

I confirmed this by demuxing with eac3to with -keepDialnorm and without it and the only one that actually worked with the Atmos channels was the -keepDialnorm. I'm not sure which header is being edited in the file itself, but editing it with a hex editor to match the one with -keepDialnorm allows the Atmos to work correctly again. However, I can't do this across other files as easily.

Is there a way to do this with eac3to or any other tool?

@jlw_4049
Yes, to preserve the Atmos data you must extract it with -keepDialnorm.

And not, I don't know a tool to restore the dialNorm.

Do either of you know if MKVToolNix has the same issue with Atmos tracks? It also has a dialnorm removal feature (Preferences/Default values/Enable dialog normalization gain removal for audio tracks).

[Encore]

what is the proper file extension for a dts-x (7.1) audio to keep the original core and object based effect. I have been extracting audio from a mkv using gMKVExtractGUI and its output was a .dts file. When I tried with TSMuxer it was same. But, as extracting tracks using eac3to was giving me 'video conversion not supported' error for -> 'eac3to source.mkv -demux' command, I used -> 'eac3to source.mkv 2: audio.dtsma' command and it gave me a audio.dtsma file without any warning or error. But my doubt here is which is the right file extension for a dts-x audio? .dtsma extension only needed for dts-hdma audio? dts.5.1 and dts-x 7.1 audios can be extracted using .dts extension without affecting the audio?

[tebasuna51]

Welcome to the Doom9's Forum Encore.

You can use always only .dts without problems.
The software must read the headers to know the kind of dts, the extension is not only used.

[Encore]

Welcome to the Doom9's Forum Encore.
Thank you..


You can use always only .dts without problems.
The software must read the headers to know the kind of dts, the extension is not only used.
If it doesn't make any change to original audio, ok. Then why eac3to demux a dts-hdma audio to .dtsma extension when u demux a bluray folder?
eg: -> eac3to 'source folder' 1) 'Destination folder'\Some-movie.*

[tebasuna51]

Just to inform you.

But the extension is not important, there are dtshd (dtsma and dtshr), dts exprees (low quality), dts-x (3D), dts 6.1 (discrete or matrixed),...
And all can have the .dts extension without problems.

Also the Master Audio Suite Encoder create .dts with a special and unic header at the begining of the file.
The decoders must read the headers to manage them.

[Encore]

Just to inform you.

But the extension is not important, there are dtshd (dtsma and dtshr), dts exprees (low quality), dts-x (3D), dts 6.1 (discrete or matrixed),...
And all can have the .dts extension without problems.

Also the Master Audio Suite Encoder create .dts with a special and unic header at the begining of the file.
The decoders must read the headers to manage them.

@tebasuna51 , thank u mate for clarifying the doubts.

On the other hand, dolby audios like TrueHD is better of .thd+ac3 while demuxing or .thd will lose its core without thd+ac3 extension?

[tebasuna51]

Extract .thd+ac3 from BD's is only usefull if you want create other BD, maybe changing video or others without changing duration.
Of course you can extract (from the m2ts) both tracks at same time N: thdpart.thd N: ac3part.ac3

If you want use other container (mkv) or make some audio edition the interleaved .thd+ac3 is useless.
BTW there are tools to restore the .thd+ac3 from the .thd and .ac3 tracks if you want create .m2ts after some edition.

[Encore]

Extract .thd+ac3 from BD's is only usefull if you want create other BD, maybe changing video or others without changing duration.
Of course you can extract (from the m2ts) both tracks at same time N: thdpart.thd N: ac3part.ac3

If you want use other container (mkv) or make some audio edition the interleaved .thd+ac3 is useless.
BTW there are tools to restore the .thd+ac3 from the .thd and .ac3 tracks if you want create .m2ts after some edition.

means, if you add a .thd+ac3 to mkv - it will not work as it was in BD or those atmos effects will be gone? And demxing a truhd atmos audio from a container other than .m2ts like .mkv not a good choice?

If so, how to make mkv from a BD with truhd atmos 7.1 audio into a mkv for back ups?

[tebasuna51]

Only m2ts container support thd+ac3 (tracks interleaved), in mkv both tracks are stored like separated tracks and must select the track you want listen.

BD players select the best for your system (maybe without thd support).

Of course you can use the mkv container without problems for BD backups, and the Atmos feature is preserved in the thd track, the ac3 track can be used in audio systems without thd support.

[maudit]

Hi, what is the syntax, if possible, to cut an audio from N time to the end?

example
-edit=1:42:04.200,-80ms #remove 80ms

cut till the end?
-edit=1:42:04.200,"WHAT HERE?" #remove till the end.

or I just specify a value greater than what I need?

[tebasuna51]

...
or I just specify a value greater than what I need?

Yes, for instance:

-edit=1:42:04.200,-9999999ms

[maudit]

Yes, for instance:

-edit=1:42:04.200,-9999999ms

ok thx, I have to cut about 5min, maybe I'll try to use "-400000ms" or more.

[Thunderbolt8]

I would like to update eac3to libflac.dll to current version 1.4.3

Here are 4 different versions of libflac.dll, though. Which one do I need?

https://www.rarewares.org/lossless.php#flac-dll

[Barough]

I would like to update eac3to libflac.dll to current version 1.4.3

Here are 4 different versions of libflac.dll, though. Which one do I need?

https://www.rarewares.org/lossless.php#flac-dll

I used this one
https://www.rarewares.org/files/lossless/flac_dll-1.4.3-x86.zip
and it worked perfect

[Encore]

Which is the best way to covert PCM 2.0 lossly to remux into a mkv. As, mkvtoolnix (tried Tsmuxer too) doesn't recognise pcm streams still I believe only way is muxing into m2ts to ass pcm audio which I don't want to save movie file as. Can I use eac3to to conver to pcm t wav and then use mkvtoolnix to remux?

[von Suppé]

Try FLAC.

[tebasuna51]

You can use eac3to to convert pcm to wav, or flac also:

eac3to input.pcm your.wav

You can obtain this message:
This might be a RAW/PCM file. Trying to figure out the details.
This will probably take a while. Please be patient...
Was not able to figure out all parameters of this RAW/PCM file.
Please specify channel, bitdepth and endian parameters via command line.

Then you need know these parameters to do the conversion, for instance:

eac3to input.pcm your.wav -override -2 -16 -big -48000

[szabi]

Hi

I did not find a way to extract core ac3 so I did encode thd 7.1ch audio by ac3-ffm with these settings:
stdout.w64 | ffmpeg -i - -c:a ac3 -b:a 640k -center_mixlev 0.707 %_.ac3
I did also try A/V Recode selecting E/AC3, ffmpeg:640.
My output is ac3 5.0ch in both cases.
I expected ac3 5.1ch output.
I do not understand where is the missing channel?

Regards
szabi

[LigH]

The ".1" channel would be a separate Low Frequency Effects channel. It is not mandatory to cover the whole frequency spectrum of the audio (especially music), it is an addition, usually produced by a sound designer for cinematic movies.

[tebasuna51]

I did not find a way to extract core ac3
Then your source is not from a m2ts BD, please put a MediaInfo of your source.
so I did encode thd 7.1ch audio by ac3-ffm with these settings:
stdout.w64 | ffmpeg -i - -c:a ac3 -b:a 640k -center_mixlev 0.707 %_.ac3
Then your source is decoded with libav dll's (2015) included in eac3to package.
And after encoded, and downmixed, to AC3 by ffmpeg.
I did also try A/V Recode selecting E/AC3, ffmpeg:640.
Now your source is decoded, downmixed and encoded by ffmpeg (easy and recommended way if you have a 5.1 audio system and don't want a more dedicated downmix (https://forum.doom9.org/showthread.php?t=181726))
My output is ac3 5.0ch in both cases.
I expected ac3 5.1ch output.
I do not understand where is the missing channel?i
Both methods preserve the LFE channel, if you obtain a 5.0 maybe your source is not a 7.1. Put the MediaInfo...

I was wrong, the automatic ffmpeg downmix have a bug at this moment and destroy the LFE channel

[szabi]

Hi

Thanks for you both for helping me to figure out.
Input was recognised by eac3to as thd 7.1ch audio.
I copy-paste the mediainfo:
Audio
Format : MLP FBA 16-ch
Format/Info : Meridian Lossless Packing FBA with 16-channel presentation
Commercial name : Dolby TrueHD with Dolby Atmos
Bit rate mode : Variable
Maximum bit rate : 5 967 kb/s
Channel(s) : 8 channels
Channel layout : L R C LFE Ls Rs Lb Rb
Sampling rate : 48.0 kHz
Frame rate : 1 200.000 FPS (40 SPF)
Compression mode : Lossless
Number of dynamic objects : 11
Bed channel count : 1 channel
Bed channel configuration : LFE
It also tells 8 channels.
My orginal intention was to keep all the 8 channel and encode it to eac 7.1ch audio.
However I was told in the eac3 7.1 encoder topic (https://forum.doom9.org/showthread.php?t=177346) it is still not possible.
So it is encoded to ac3 but I lost an other channel too.

Regards
szabi

[tebasuna51]

You are rigth and the bug is when ffmpeg downmix 7.1 automatically:
Input #0, truehd, from 'C:\Mis_Doc\Audio71\Core\thd\CoreAT.thd':
Stream #0:0: Audio: truehd, 48000 Hz, 7.1, s32 (24 bit)
Output #0, ac3, to 'C:\tmp\CoreAT.thd_.ac3':
Stream #0:0: Audio: ac3, 48000 Hz, 5.0(side), fltp, 640 kb/s

You can use the eac3to downmix with:

stdout.w64 -down6 | ffmpeg -i - -c:a ac3 -b:a 640k -center_mixlev 0.707 %_.ac3

or a better ffmpeg downmix, instead the automatic, explained in the mentioned thread (https://forum.doom9.org/showthread.php?t=181726).

[szabi]

I actually do not understand from the mentioned thread what I should do now.
I downloaded the newest ffmpeg now but no luck with BtbN or gyan version, both has the issue.
Because formarly this issue did not exist I assume there is an older version ffmpeg which works properly.
Might you know which is the last correct ffmpeg?

[richardpl]

To even be able to help someone, first needs exact command someone tried with full output of that command with any log message and errors it can give.

[tebasuna51]

To even be able to help someone, first needs exact command someone tried with full output of that command with any log message and errors it can give.

The bug is clear, tested with any source (wav,w64,thd,...) 7.1:

ffmpeg -i 7.1.any -acodec ac3 -ab 640k output.ac3

without warnings or errors the output is 5.0 like show the previous post.

[tebasuna51]

I actually do not understand from the mentioned thread what I should do now.

A recommended downmix 7.1 -> 5.1 to preserve the full volume in the front channels and mix the 4 surround channels in only 2 without clips is:
ffmpeg -i input71.thd -filter_complex "asplit [f][s]; [f] pan=3.1|c0=c0|c1=c1|c2=c2|c3=c3 [r]; [s] pan=stereo|c0=0.5*c4+0.5*c6|c1=0.5*c5+0.5*c7, compand=attacks=0:decays=0:points=-90/-84|-8/-2|-6/-1|-0/-0.1, aformat=channel_layouts=stereo [d]; [r][d] amerge [a]" -map "[a]" -acodec ac3 -center_mixlev 0.707 -ab 640k output.ac3

[szabi]

I copy-paste it to command line, the result:
https://abload.de/img/ffmpegjaipx.png
What else I need to do?

[LigH]

Looks like there is a problem parsing that command line... check twice if there is an unexpected wrap, maybe prefer writing it in a batch file (which you can edit easier later).

[d4rk3]

I copy-paste it to command line, the result:
https://abload.de/img/ffmpegjaipx.png
What else I need to do?

Try this:

ffmpeg -i "input71.thd" -filter_complex "asplit [f][s]; [f] pan=3.1|c0=c0|c1=c1|c2=c2|c3=c3 [r]; [s] 'pan=stereo|c0=0.5*c4+0.5*c6|c1=0.5*c5+0.5*c7', 'compand=attacks=0:decays=0:points=-90/-84|-8/-2|-6/-1|-0/-0.1', aformat=channel_layouts=stereo [d]; [r][d] amerge [a]" -map "[a]" -acodec "ac3" -center_mixlev "0.707" -ab "640k" "output.ac3"

[szabi]

I did try your command line but the result is the same: don't found .exe

[tebasuna51]

Copy it to a X.bat file, edit it to use your path's to ffmpeg.exe, input and output files and execute the X.bat

[szabi]

Hi

All of you are right copy to bat has worked.
So presently this is the best free technique to encore 7.1ch audio to 5.1ch ac3?
Because still no free 7.1ch encoder to eac3, how can I encode the input to 5.1ch eac3?
Is it worth to use eac3 instead of ac3?
I still interested why it did not work over UsEac3To? Also it would be nice to have this option in UsEac3To

It might be useful to someone later so bat file I used:
"C:\My Program Files\audio-encoders\ffmpeg.exe" -i "E:\1\engaudio.thd" -filter_complex "asplit [f][s]; [f] pan=3.1|c0=c0|c1=c1|c2=c2|c3=c3 [r]; [s] pan=stereo|c0=0.5*c4+0.5*c6|c1=0.5*c5+0.5*c7, compand=attacks=0:decays=0:points=-90/-84|-8/-2|-6/-1|-0/-0.1, aformat=channel_layouts=stereo [d]; [r][d] amerge [a]" -map "[a]" -acodec ac3 -center_mixlev 0.707 -ab 640k "E:\1\engaudio.ac3"

Last but not least thanks for your help.

Regards
szabi

[tebasuna51]

So presently this is the best free technique to encode 7.1ch audio to 5.1ch ac3?
I think so.
I still interested why it did not work over UsEac3To? Also it would be nice to have this option in UsEac3To

It is already present, see the attached image.
When select the bitrate the E prefix encode to EAC3, of course with better quality than AC3.

When the source is a container (a m2ts for instance) remember select the track with map:X in the Audio Recode window.
(X for ffmpeg is 1 less, normally, than for eac3to)

[szabi]

Hi

With your guidance I could do it over UsEac3To.
I use ffmpeg version N-112483-gd799ad2404-20231020, with UsEac3To v1.3.3.
The FFMPEG functions parameters a bit different when I select 71-51.
https://abload.de/img/ffmpeg-cmdgii2p.png

Is this different an issue? Should I modify it to your parameters?
points=-90/-84|-10/-4|-6/-2|-0/-0.3
points=-90/-84|-8/-2|-6/-1|-0/-0.1

Also during encode it was written:
Input #0, truehd, from 'E:\1\engaudio.thd':
Duration: N/A, start: 0.000000, bitrate: N/A
Stream #0:0: Audio: truehd (Dolby TrueHD + Dolby Atmos), 48000 Hz, 7.1, s32 (24 bit)
[out#0/eac3 @ 0000024a207781c0] Codec AVOption center_mixlev (Center Mix Level) has not been used for any stream. The most likely reason is either wrong type (e.g. a video option with no video streams) or that it is a private option of some encoder which was not actually used for any stream.
Stream mapping:
Stream #0:0 (truehd) -> asplit:default
amerge:default -> Stream #0:0 (eac3)
Press [q] to stop, [?] for help
[Parsed_amerge_5 @ 0000024a20804180] Input channel layouts overlap: output layout will be determined by the number of distinct input channels
[Parsed_pan_1 @ 0000024a207778c0] Pure channel mapping detected: 0 1 2 3
Output #0, eac3, to 'E:\1\engaudio.thd_.eac3':
Metadata:
encoder : Lavf60.15.100
Stream #0:0: Audio: eac3, 48000 Hz, 5.1, fltp, 800 kb/s
Metadata:
encoder : Lavc60.30.102 eac3

Is the two line in red a problematic message?

And once more I just repeat myself many thanks for your help.

Regards
szabi

[tebasuna51]

The FFMPEG functions parameters a bit different when I select 71-51.
Select yourself the desired compand curve (https://forum.doom9.org/showthread.php?p=1926777#post1926777) to manage high volumes.
If the 4 surround channels have high volume we can't obtain the same volume with only 2 speakers but is only in a few moments in the movie (explosions, Rex roaring, ...). Most the time the 2 surround speakers of 5.1 can supply the same volume than the 4 of the 7.1.
Is the two line in red a problematic message?
No problem, forget them.

[szabi]

Select yourself the desired compand curve (https://forum.doom9.org/showthread.php?p=1926777#post1926777) to manage high volumes.
If the 4 surround channels have high volume we can't obtain the same volume with only 2 speakers but is only in a few moments in the movie (explosions, Rex roaring, ...). Most the time the 2 surround speakers of 5.1 can supply the same volume than the 4 of the 7.1.

No problem, forget them.

Hi

I might understand what u say but definetly cannot decide the default or your parameters should be selected.

EDIT:
This is the point when I am back to my initial 8 years old question about 7.1ch EAC3 output is desired.

Regards
szabi

[tebasuna51]

I might understand what u say but definetly cannot decide the default or your parameters should be selected.
Both are ok, and sugested by me, with little differences.

This is the point when I am back to my initial 8 years old question about 7.1ch EAC3 output is desired.
I'm not interested in 7.1 audio, to obtain sound 2D the 5.1 is more than enough.
But maybe 5.1.2 to obtain 3D sound from Atmos sources can be interesting.

[szabi]

What to do if no free 7.1ch eac3 encoder. It is what it is.
Once more thanks for help. :)