Proper way to FFMPEG DTS-MA 7.1 to AC3 5.1?

[tormento]

I have googled lot of sites but didn't find answer.

Is possible to use ffmpeg to convert a DTS-MA 7.1 file to AC3 5.1 (640kbps) with proper channel mapping and mixdown?

Eac3to gives me error and ffmpeg seems my only viable way...

[microchip8]

yes, just use -ac 6 in ffmpeg and it does its auto-magic. I use it all the time and the result is always correct (tested on my surround system)

[tormento]

Originally Posted by froggy1

yes, just use -ac 6 in ffmpeg and it does its auto-magic. I use it all the time and the result is always correct (tested on my surround system)

Great, thanks. I tried also to find a gui that process audio files only but nothing found. What a pity.

[tebasuna51]

ffmpeg and eac3to (-down6) can do the job.

But when downmix the overall volume can be reduced, to avoid clip when SL,BL and SR,BR are mixed.
I can recommend a special AviSynth downmix to preserve the full volume in FL,FR,FC,LFE channels.
In short something like:

# Downmix 7.1 to 5.1
function 71to51(clip a) {
  flr = Getchannel(a, 1, 2, 3, 4)
  blr = Getchannel(a, 5, 6)
  slr = Getchannel(a, 7, 8)
  sur = MixAudio(blr, slr, 1.0, 1.0).SoftClipperFromAudX(0.0)
return mergechannels(flr,sur)
}

That it is available in MeGUI or BeHappy.
For instance in MeGUI with a DTS-HD source:

[tormento]

Originally Posted by tebasuna51

# Downmix 7.1 to 5.1
function 71to51(clip a) {
  flr = Getchannel(a, 1, 2, 3, 4)
  blr = Getchannel(a, 5, 6)
  slr = Getchannel(a, 7, 8)
  sur = MixAudio(blr, slr, 1.0, 1.0).SoftClipperFromAudX(0.0)
return mergechannels(flr,sur)
}

Tell me more about this, please.

I never used AVS to encode/decode audio.

[tebasuna51]

That avs function split the channels in:

flr = Getchannel(a, 1, 2, 3, 4) # channels FL, FR, FC, LFE
blr = Getchannel(a, 5, 6) # channels BL, BR
slr = Getchannel(a, 7, 8) # channels SL, SR

Now mix channels BL with SL, BR with SR

sur = MixAudio(blr, slr, 1.0, 1.0)

The mix is at full volume (1.0) and can produce clip if sum of volumes is greater than 1.0 (float samples are used at this point)
And apply a plugin (Audiolimiter.dll) to reduce volume only if clip is detected, to limit max volume at 1.0:

sur = sur.SoftClipperFromAudX(0.0)

and finally return the 6 channels without changes in 'flr'

return mergechannels(flr,sur)

Then most the time the volume in 'sur' is the same than in the 4 surround channels, and the global balance between front and surround is preserved.
Only in clip points the balance is lost and 'sur' (two speakers) can't produce the same volume than BL,BR,SL,SR (four speakers).

The standard way to mix 7.1 -> 5.1 is lower the volume of ALL channels to prevent clips and you have correct balance in all the movie but at half volume.

[tebasuna51]

OTHER QUESTIONS:

- That MeGUI method work also for 6.1 sources (other fuction 61to51 is selected) and DTS-HR 96 KHz (Change to 48000, than do nothing if it is already 48000).

- The LWLibavAudioSource can decode many audio sources (DTS-HD, TrueHD, FLAC, AAC, ...), and also the first audio track of containers like m2ts or mkv.

- You can include other parameters to ffmpeg encoder in Extra Tab, for instance I recommend:

-center_mixlev 0.707

to override the ffmpeg default 0.594604, to low for me, used if decoder need downmix to stereo that 5.1.

[tormento]

Originally Posted by tebasuna51

That avs function split the channels in

And how can AviSynth read a DTS?

I really miss all the audio part.

[tebasuna51]

The same than video, with a proper spliter/decoder.
The plugin LSMASHSource.dll (also ffms2.dll) have functions:

LWLibavVideoSource("inputfile",...)
LWLibavAudioSource("inputfile",...)
...

It convert many sources to uncompressed PCM samples than can be managed inside AviSynth with internal Audio processing filters and with external Audio filters

You can see the docs for full explain.

[SeeMoreDigital]

Originally Posted by tormento

Is possible to use ffmpeg to convert a DTS-MA 7.1 file to AC3 5.1 (640kbps) with proper channel mapping and mixdown?

Is there any particular reason why you require Dolby Digital (AC3) audio stream and can't use the DTS core?


Cheers

[tormento]

Originally Posted by tebasuna51

You can see the docs for full explain.

Thanks.

[tebasuna51]

There are two generic reasons:

-Compatibility. Not all players support DTS decode.

- Size. 42% size of a core 1509 Kb/s.

[tormento]

Originally Posted by tebasuna51

There are two generic reasons:
-Compatibility. Not all players support DTS decode.
- Size. 42% size of a core 1509 Kb/s.

Both are valid to me.

[SeeMoreDigital]

Okay...

Have you considered using the 5.1 DTS core as a source to generate your 5.1 AC3 encode?

[microchip8]

Originally Posted by SeeMoreDigital

Okay...

Have you considered using the 5.1 DTS core as a source to generate your 5.1 AC3 encode?

Compressing compressed audio... compressing compressed audio... and stuff like that

[SeeMoreDigital]

Originally Posted by froggy1

Compressing compressed audio... compressing compressed audio... and stuff like that

Yes we all know this, but....

Unless tormento has a well spec'd surround sound set-up (which he can't have or he'd be keeping the original DTS-MA stream or using the DTS core), he's unlikely to hear that much of a difference!

[Vicio]

Originally Posted by tebasuna51

OTHER QUESTIONS:

- That MeGUI method work also for 6.1 sources (other fuction 61to51 is selected) and DTS-HR 96 KHz (Change to 48000, than do nothing if it is already 48000).

- The LWLibavAudioSource can decode many audio sources (DTS-HD, TrueHD, FLAC, AAC, ...), and also the first audio track of containers like m2ts or mkv.

- You can include other parameters to ffmpeg encoder in Extra Tab, for instance I recommend:

-center_mixlev 0.707

to override the ffmpeg default 0.594604, to low for me, used if decoder need downmix to stereo that 5.1.

Hello, when should the command line (-center_mixlev 0.707) be used?
Only for 6.1 / 5.1 to 2.0 audio conversions or should it also be used for 2.0 to 2.0 audio conversions?
I ask this because I went to change the speed of a DDP audio (EAC3) with eac3to.exe the conversion was carried out, but however the command was not applied.

[tebasuna51]

Originally Posted by Vicio

Hello, when should the command line (-center_mixlev 0.707) be used?
Only for 6.1 / 5.1 to 2.0 audio conversions or should it also be used for 2.0 to 2.0 audio conversions?

1) The center_mixlevel is not a field in eac3 bsi header then ffmpeg say than can't be used. Correct.
My template is for ac3 encode, you can change it for eac3 but you can delete also that parameter.

2) In ac3 2.0 bsi header also don't exist the field center_mixlevel, you can also delete it. My template is for multichannel because for 2.0 always recommend recode to AAC.

BTW ffmpeg yelow warnings aren't a problem, are only info.

[tebasuna51]

To change duration of audio tracks you can use UsEac3to and ffmpeg without eac3to.

Sometimes eac3to can't decode the source track or you want preserve the pitch of the track (eac3to change the pitch when you change the audio duration).

See the image. You want change the EAC3 english track.

1) Use the 'More parameters' -> F-FFMPEG -> Add or write directly in 'COMMAND LINE PARAMETERS' the ffmpeg audio filter:
-af "atempo=1.001"

with:

 fps's video     TimeStretch ffmpeg
---------------- -------------------
23.976 -> 25.000   1.042708  speedup
23.976 -> 24.000   1.001
24.000 -> 25.000   1.041667
24.000 -> 23.976   0.999001
25.000 -> 24.000   0.96
25.000 -> 23.976   0.959041  slowdown

2) Open the 'A/V Recode' window

3) Select the track to modify. Remember ffmpeg start by track '0' then the eac3to track 3: is the map=0:2 for ffmpeg

4) 'Add to DEC' the ffmpeg filter included in the 'COMMAND LINE PARAMETERS'

5) Select the output format and bitrate/quality: AC3, EAC3, AAC (recommended for 2.0), ...

6) 'Run' or 'EnQueue'

- Like this thread is for Downmix 7.1 to 5.1 the method can be used also for that using the ffmpeg filter in 1):

-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=0oints=-90/-84|-10/-4|-6/-2|-0/-0.3, aformat=channel_layouts=stereo [d]; [r][d] amerge [a]" -map "[a]"

Remember, the is

:p

and 2) to 6) steps the same.

[Vicio]

Originally Posted by tebasuna51

To change duration of audio tracks you can use UsEac3to and ffmpeg without eac3to.

Thank you very much, the tips are of great help.
In fact, when it comes to handling EAC3 audios I really prefer BeHappy, I asked the question because I was surprised to see that in UsEac3to it was possible to change the speed of this audio format as well.

When I saw that he did not apply the command “-center_mixlev 0.707” to the EAC3 audio, I was in doubt if the command was really necessary for 2.0 audios
But now with your explanation I know that it is intended for multichannel only.

Greetings from Brazil.