AudioMuxer - 2nd step necessary to decode strange Dolby AC3 stream

[ultradj83]

Dears,
I'm extracting a AC3 stream from MKV, to be edited in Audition.

This stream has strange behaviors.
It's the usual 2.0 384kbps AC3, as Mediainfo confirms.

By the way, opening the MKV directly in Audition a 5.1 track is decoded, with LFE channel 99% blank, some voices missing, by the way L R C and Lr Rr channels appear to be normal; no idea where this strange 5.1 comes from, maybe Audition is decoding Pro Logic?

Different behavior with Audiomuxer.
Attempting to convert the audio track from the MKV generates a 2.0 WAV file, but it's totally corrupted and, at the end of the usual decoding step, I have this:

[Attachment 69524 - Click to enlarge]
No idea what's happening.
Moreover, by extracting the AC3 track from MKV, I have a coherent 2.0 AC3 file, by the way also in this way I see the 2nd necessary step and the resulting file is absolutely too loud, maybe something is failing to handle AC3 diagnorm?

Any suggestions?

Regards

[ProWo]

Could be a 5.1 ac3 stream with leading a few 2ch frames.
This can happen with TV recordings.

[ultradj83]

Could be a 5.1 ac3 stream with leading a few 2ch frames.
This can happen with TV recordings.

Maybe this was originally delivered to TV.
But what you mean exactly?

Furthermore, this is the EAC3to output:

a02 Extracting audio track number 2...
a02 Removing AC3 dialog normalization...
a02 Decoding with libav/ffmpeg...
a02 Reducing depth from 64 to 24 bits...
a02 Writing WAV...
a02 Creating file "\\ULTRANAS\Download\output.wav"...
a02 This track is not clean.
a02 Caution: The WAV file is bigger than 4GB.
a02 Some WAV readers might not be able to handle this file correctly.
a02 Audio has a gap of 32ms at playtime 0:40:12.
a02 Audio has a gap of 64ms at playtime 1:11:48.
a02 Starting 2nd pass...
a02 Extracting audio track number 2...
a02 Removing AC3 dialog normalization...
a02 Decoding with libav/ffmpeg...
a02 Reducing depth from 64 to 24 bits...
a02 Writing WAV...
a02 Realizing RAW/PCM gaps...
a02 Creating file "\\ULTRANAS\Download\output.wav"...
a02 Caution: The WAV file is bigger than 4GB.
a02 Some WAV readers might not be able to handle this file correctly.
Video track 1 contains 194375 frames.
eac3to processing took 7 minutes, 54 seconds.
Done.

For sure file is not clean, parameters are read correctly, but output file is totally corrupted, sounds like wrong sample rate...


Regards.

[ProWo]

Extract the ac3 and post it here, so we can take a look.

[ultradj83]

Extract the ac3 and post it here, so we can take a look.

Dear,
here it is: https://we.tl/t-iNObH9rQ4P .

Always a pleasure to learn something.

Thanks in advance.


Regards

[noemi7]

I used dBpoweramp to converted it to wav. Looks strange
[Attachment 69528 - Click to enlarge]

[ultradj83]

I used dBpoweramp to converted it to wav. Looks strange
[Attachment 69528 - Click to enlarge]

Yea, by the way looks better than any attempts of mine. Plays fine?
How did you do that?

I installed dbpoweramp, which is using LAV filters + Directshow to decode AC3, and it's decoding the 2.0 AC3 to 5.1 WAV:

[Attachment 69529 - Click to enlarge]
It is inventing 5.1 as well as Adobe Audition, and again the LFE is pretty blank:

[Attachment 69530 - Click to enlarge]
Audition uses Windows Media Foundation to decode, which might be relaying on Directshow, too.

In the meanwhile, if I directly open the AC3 track I uploaded, which is simply extracted from the MKV by MKVextract, I have readable content but L channel is louder then R channel:

[Attachment 69531 - Click to enlarge]

It's becoming stunning

Still can only imagine this is the result of any matrix encoding, can't imagine any other reasons...


Regards

[ProWo]

As i sayed, ac3 5.1 with leading 2ch samples (9).
Use clever FFmpeg-GUI to patch such ac3 audio files.


[Attachment 69540 - Click to enlarge]

[Attachment 69541 - Click to enlarge]

[Attachment 69542 - Click to enlarge]

[Attachment 69543 - Click to enlarge]

[ultradj83]

As i sayed, ac3 5.1 with leading 2ch samples (9).
Use clever FFmpeg-GUI to patch such ac3 audio files.


[Attachment 69543 - Click to enlarge]

Thanks for suggestion.
So you think it's 2.0 AC3 with Pro Logic?
By the way this is exactly the same output I have with Audition, with LFE pretty absent...

[Attachment 69541 - Click to enlarge]

Moreover I couldn't find FFmpeg-GUI 3.0.9.01, and on FFmpeg-GUI 3.0.9 I don't have the AC3 5.1 Patch button.

Then, scanning file, I have thousands of such errors:

[Attachment 69544 - Click to enlarge]


Regards

[ProWo]

So you think it's 2.0 AC3 with Pro Logic?

No, it's a 5.1 ac3 with 2ch ac3 samples at the beginning.

Moreover I couldn't find FFmpeg-GUI 3.0.9.01, and on FFmpeg-GUI 3.0.9 I don't have the AC3 5.1 Patch button.

The AC3 5.1 Patch button is there, but is shown only with ac3 5.1 sources. BTW 3.0.9.01 is online now.

Then, scanning file, I have thousands of such errors:

If the video works, forget the errors.

[ultradj83]

No, it's a 5.1 ac3 with 2ch ac3 samples at the beginning.

Got it definitely, thanks.

The AC3 5.1 Patch button is there, but is shown only with ac3 5.1 sources. BTW 3.0.9.01 is online now.

Ok, sorry but at this point I might have lost a bit.
If it is already 5.1, why should I use the AC3 5.1 Patch?
What is it exactly doing?
Is it padding broken/missing samples or somethin else?

If the video works, forget the errors.

In your opinion, the LFE pretty blank, is an error or a strange choice of production?


Regards

[ProWo]

Ok, sorry but at this point I might have lost a bit.
If it is already 5.1, why should I use the AC3 5.1 Patch?
What is it exactly doing?
Is it padding broken/missing samples or somethin else?

Is used ONLY for audio streams reported by ffmpeg as ac3 5.1, but reported by mediainfo and others as ac3 stereo (2ch).
It replaces the wrong leading 2ch headers with 5.1 headers.
This problem happens often by bad cuts of TV recordings.

In your opinion, the LFE pretty blank, is an error or a strange choice of production?

It's not an error, it's the original audio stream.

[Cornucopia]

In your opinion, the LFE pretty blank, is an error or a strange choice of production?

It is a NORMAL choice of production. You want the LFE to be as loud as your main signal? That would shake your house apart! (only partly exaggerating)
Most balanced mixes have a quite small amount of LFE - remember LFE properly used should only include Low Frequency Effects, not standard LF or subwoofer-intended sound.
The choice of going to main vs subwoofer speakers, etc is determined by the playback decode software/device, based on the room's equipment.

Scott

[ultradj83]

In your opinion, the LFE pretty blank, is an error or a strange choice of production?

It is a NORMAL choice of production. You want the LFE to be as loud as your main signal? That would shake your house apart! (only partly exaggerating)
Most balanced mixes have a quite small amount of LFE - remember LFE properly used should only include Low Frequency Effects, not standard LF or subwoofer-intended sound.
The choice of going to main vs subwoofer speakers, etc is determined by the playback decode software/device, based on the room's equipment.

Scott

Right.
But mine is totally silent...
I rebuilt the LFE by moving all channels below 80Hz, then applying -12dB gain.

Regards.

[Cornucopia]

Again, that is not what LFE channel is for. What should be in LFE should be unique to the LFE channel, a supplement to the main channels. If you are deriving your LFE from other channels, all you are doing is pumping up the bass, which is not necessary with proper home theater bass management. If your original truly had nothing in LFE (many fine productions do not), it cannot and should not be synthesized, and doing so regardless is counter to "high fidelity". If you know for a fact that it was supposed to have something when yours didn't, then it is more a matter of finding a better source copy or of using a better decoding tool.

Scott