Eclipsa Audio - Spatial Audio Format by Samsung and Google

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

LennonCobain

1K Club - QQ Shooting Star
QQ Supporter
Joined
Feb 24, 2014
Messages
1,796
Location
New England
I found this repository to better understand what is all about:
https://github.com/AOMediaCodec/iamf-tools
More IAMF Spatial audio info in the links below
https://aomedia.org/iamf/
https://aomediacodec.github.io/iamf/
Now also in ffmpeg:
https://ffmpeg.org/
https://ffmpeg.org/ffmpeg.html#:~:tex... Audio"The libavformat library can now read and write IAMF (Immersive Audio) files. The ffmpeg CLI tool can configure IAMF structure with the new -stream_group option." FFMPEG
Note "IAMF" is the "old" name for what is now called "eclipsa"

YouTube will support it (we'll have to see what that means. HDMI from your computer to your Samsung TV/Soundbar device? Or maybe just headphones?
 
More IAMF Spatial audio info in the links below
https://aomedia.org/iamf/
https://aomediacodec.github.io/iamf/
Now also in ffmpeg:
https://ffmpeg.org/
https://ffmpeg.org/ffmpeg.html#:~:tex... Audio"The libavformat library can now read and write IAMF (Immersive Audio) files. The ffmpeg CLI tool can configure IAMF structure with the new -stream_group option." FFMPEG
Note "IAMF" is the "old" name for what is now called "eclipsa"

YouTube will support it (we'll have to see what that means. HDMI from your computer to your Samsung TV/Soundbar device? Or maybe just headphones?
Every smart TV and streaming device has access to the YouTube app. Presumably it means that app will now stream in surround.

If it means I can stream music in surround on my computer though? I’m up for it
 
Last edited:
For those of you that are into the gory details, I have built the decoder and the tools (includes an encoder) on windows 11.

In theory you don't need the tools, given ffmpeg can do the encode and merge.

Building the tools was painful and took most of the morning, but I got it done.

Building the decoder was super easy in VS2022, as a VS2022 .sln file was provided.

Next I will be doing some encodes and decodes. They have examples for up to 7.1.4 and a test file for 9.1.6. So far I haven't seen anything about objects.

It would be nice to know the details on YouTube support, as a supported player, but I haven't seen anything yet about that either.
 
For those of you that are into the gory details, I have built the decoder and the tools (includes an encoder) on windows 11.

In theory you don't need the tools, given ffmpeg can do the encode and merge.

Building the tools was painful and took most of the morning, but I got it done.

Building the decoder was super easy in VS2022, as a VS2022 .sln file was provided.

Next I will be doing some encodes and decodes. They have examples for up to 7.1.4 and a test file for 9.1.6. So far I haven't seen anything about objects.

It would be nice to know the details on YouTube support, as a supported player, but I haven't seen anything yet about that either.
So, so far it is all about downmix and upmix ?
I saw some ambisonic in the specification, at least it would be cool if YouTube supports a higher order than one.
YouTube already supports opus, you can upload a 5.1 YouTube video, you can see the file in the YouTube manifest, but there are no players for it.
https://peachymango.org/Surround
 
So, so far it is all about downmix and upmix ?
I saw some ambisonic in the specification, at least it would be cool if YouTube supports a higher order than one.
YouTube already supports opus, you can upload a 5.1 YouTube video, you can see the file in the YouTube manifest, but there are no players for it.
https://peachymango.org/Surround
No, it just does (some of) what atoms and dts x does. A way to get immersive (and 5.1) surround to the end user. Downmix yes, upmix no. Can do 1st and 3rd order ambisonics, but also discrete speaker channels. I’m still reading.

I have encoded and decoded 7.1.4. Also decoded to binaural, but with the default inputs it didn’t “spatialize” it for my ears. However it looks like you can use BRIR / HRIR of your choice so that is a whole area of research.

I have also found a VST eco system for producing content in Reaper. I’ll post the link when I’m next at my computer.
 
No, it just does (some of) what atoms and dts x does. A way to get immersive (and 5.1) surround to the end user. Downmix yes, upmix no. Can do 1st and 3rd order ambisonics, but also discrete speaker channels. I’m still reading.

I have encoded and decoded 7.1.4. Also decoded to binaural, but with the default inputs it didn’t “spatialize” it for my ears. However it looks like you can use BRIR / HRIR of your choice so that is a whole area of research.

I have also found a VST eco system for producing content in Reaper. I’ll post the link when I’m next at my computer.
https://ear-production-suite.ebu.io/
 
The pessimistic side of me wants to say "Samsung wants to lose another format war (nobody cares about HDR10+). But Google invented VP9 and it stuck, so this time may be different.

If that write-up is an accurate summary of the formats intentions, it's all a really fancy way to say it's an AI up-mixer that allows the end user to tweak levels. "More drums please, put more guitar in the overheads, turn off Cris Collinsworth's/Tony Romo's mic, etc".
 
Re the binaural decoder, it turns out in the windows build doesn't include it. Errr. However, there is a web demo that will do it here: https://aomediacodec.github.io/iamf-tools/web_demo/

So, short of buying 2025 Samsung gear, this is the only way to kick the tires on IAMF / Eclipsa.

I took a 7.1.4 up-remix I had already done of America - Donkey Jaw, and converted it to IAMF with FFMpeg, then decoded to Binaural with the web demo. It's OK I guess, spatially (I'm comparing to Smyth Realizer A16 with and 7.1.4 Genelecs). Sounds a little tinny, but then they say they are still working on the Binaural stuff. I guess the Sony MDR-7506s I listened to the Binaural are not in the same class as the Sennheiser HD-800 S cans I use on the A16 either (an also not personalized like the Sennheiser's either).
 
The pessimistic side of me wants to say "Samsung wants to lose another format war (nobody cares about HDR10+). But Google invented VP9 and it stuck, so this time may be different.

If that write-up is an accurate summary of the formats intentions, it's all a really fancy way to say it's an AI up-mixer that allows the end user to tweak levels. "More drums please, put more guitar in the overheads, turn off Cris Collinsworth's/Tony Romo's mic, etc".
Like VP9, it may only work if it becomes part of the web browser formats. Web browsers like Firefox need license free codecs. Outside that type of space, it does not seem that license free codecs get any market traction.
 
Back
Top