MMH - New Atmos Decoder (beta) discussion

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I have decoded Truehd atmos MLP to 16 channels into .w64 file using your atmos decoding. Then mix the content into channel C of the w64 file. From the new w64 file, I created an atmos file and encoded it to truehd. After completion, I compared the bitrate of the new and old TrueHD and saw a difference of about 300kbps. So do you know any tips or advice to recommend that the bit rate not be reduced much? Thank
1703304965094.png
 
Last edited:
Edit: (After seeing the attached image):

300kbps is nothing to worry about when the source is over 6,500kbps. Its probably the object metadata that gets removed in the decode to 16 channels.

You could check if the merged MCH file is 16 bit. Ffmpeg defaults to 16bit but not sure if that applies to the channel merging? but zi doubt that’s the issue.
 
Last edited:
@Tibblouder If you don’t playback on a 9.1.6 system: Since the original mix has 11 objects I’d think it was a a 7.1.4 mix, so you could probably decode to 7.1.4 (12 channel), do your centre channel edit, then re-encode to 7.1.4 instead of 9.1.6.
 
or maybe lost data during the process of creating a master file? Can you reveal how the master file was created?
 
Last edited:
or maybe lost data during the process of creating a master file? Can you reveal how the master file was created?

It will be the missing metadata since MMH is creating master files with the multichannel wav converted to CAF. Thats a lossless conversion to CAF with stationary object metadata.
 
This may be the cause of the bit rate decrease.
I just did a little comparison.
I decode atmos audio with eac3to and only get channel C, continue decoding atmos with helper and get channel C to compare with channel C just decoded with eac3to and see that:
There are 4 or 5 small pieces of sound lost in C-helper compared to C-eac3to.
To compare, I opened 2 C channels on soundforge, then invert 1 audio, copy the invert audio and mix it into the remaining audio. If channel C was completely the same at first, then after mixing, the entire sound would be lost, but no. , there were 4 or 5 places (or more) left in the sound, I marked it again to see the difference and it was true, your decoder may have lost some data, not lost data in the step create atmos file.
 
The Atmos Helper uses the Dolby Reference Player to decode Atmos. ‘Reference Player’ is certified by Dolby.

What is eac3to using? I thought it could only decode the embedded AC3 stream of a blu-ray Atmos stream? (And maybe DDP). As far as I know it can’t decode TrueHD Atmos, it can only decode the 7.1 bed. How many channels are you getting when you decode with using eac3to?
 
I was wrong, if I decode with eac3to, I only get 8 channels. So channel C will be a little different.
So let me ask, what mechanism or software do you use to create the atmos master file?
 
So let me ask, what mechanism or software do you use to create the atmos master file?

I wrote the code after looking at what data was in the Atmos master file sets. With a little help from another QQ member we were able to reproduce those sets.

Remember MMH just allows multichannel interleaved mixed files to be encoded by the Dolby Encoder Engine (The encoder rejects MCH files on their own). To do that MMH creates master file sets. The input channels are defined as fixed objects at the standard Dolby speaker positions in the master sets used in the encode.

For object based mixes users need to purchase professional Atmos mixing and rendering software. Nuendo Cubase Pro was what I used to do my first Atmos mixing project and that’s where I worked out what the Atmos master file sets were.
 
The bitrate loss will be at decoding or master file creation.
Hopefully in the near future there will be a new version that when re-encoding atmos, the bit rate will be less reduced than currently, thank you!
 
Last edited:
The Atmos Helper uses the Dolby Reference Player to decode Atmos. ‘Reference Player’ is certified by Dolby.

What is eac3to using? I thought it could only decode the embedded AC3 stream of a blu-ray Atmos stream? (And maybe DDP). As far as I know it can’t decode TrueHD Atmos, it can only decode the 7.1 bed. How many channels are you getting when you decode with using eac3to?
You are correct. Eac3to can't decode any form of Atmos meta-data.

It can decode a lossless 8-channel TrueHD stream without Atmos. It can even create a 6-channel lossy Dolby Digital 'core' encode from the lossless 8-channel Dolby TrueHD stream. Which is very useful if you want to create .m2ts muxes.

That being said, eac3to has recently acquired some new developers. So who knows what treats await ;)
 
No attachment.

Hopefully in the near future there will be a new version that when re-encoding atmos, the bit rate will be less reduced than currently,

There will be no other new version. I’ve explained the original object metadata won’t be in the reencode.

No one would hear the 320kbps difference at 6500kbps anyway.
 
Hi Garry,

Happy new year.

I think I've picked up a bug in the MMH Atmos decoder... If you batch process a whole bunch of folders and there are two or more files in the queue with the same filename, the decoding process stalls when it gets to the second file. I assume this is due to temp file naming or something?
 
I think I've picked up a bug in the MMH Atmos decoder

Happy new year!

Thanks for reporting. The MMH Atmos Helper’s decoder uses a temporary folder, so I think you are correct that if two separate album folders have files with same names that could be a problem if the two were being processed at same time in the temp folder.

I’ll take a look a get a fix later today.

I’ve been meaning to do an official release for a while, a few fixes and changes ready to roll. (Released in betas only).
 
For a test, I tried to decode the audio track (THDA) from the movie Twister. I chose 7.1.4 layout. Each time, your program only decoded 58 minutes in both "multichannel wav" and "mono wavs" modes.
Have you encountered such a problem before?
I send greetings to you.
 
For a test, I tried to decode the audio track (THDA) from the movie Twister. I chose 7.1.4 layout. Each time, your program only decoded 58 minutes in both "multichannel wav" and "mono wavs" modes.
Have you encountered such a problem before?
I send greetings to you.
Is that the only one you have tried, and it didn't work? I have successfully done a decode longer than that, but I believe the original format was different. Do you have another movie you can try?
 
Each time, your program only decoded 58 minutes in both "multichannel wav" and "mono wavs" modes.

Have you got plenty of spare space on your Windows temp drive?
When the decoder command window opens, does the progress reach 100% before it closes?
 
Back
Top