search

ebu / ear-production-suite

The EAR Production Suite is a set of VST® plugins and tools for producing immersive and personalizable audio content suitable for any Next Generation Audio codec. It is based on the Audio Definition Model (ITU-R BS.2076) and the ITU ADM Renderer (ITU-R BS.2127) and enables monitoring on any ITU-R BS.2051 loudspeaker configuration.
https://ear-production-suite.ebu.io/
GNU General Public License v3.0
102 stars 19 forks source link

YOUR ADM BWF EXPORT FUNCTION FOR REAPER IS NO LONGER WORKING CORRECTLY #51

Closed mpsrikar closed 3 years ago

mpsrikar commented 3 years ago

I downloaded your suite 15 days ago.... for reaper, windows 10. I completed making an object-based audio mix using EAR production suite just yesterday. Now I exported it in Adm bwf 38 channels wav. I decided to encode it to Dolby Atmos. But the Dolby Atmos conversion tool is not recognizing your exported Adm bwf format of EAR production suite from reaper. Dolby Atmos conversion tool is not able to recognize it and says track sample rate and ADMs sample rate mismatch error. On the other hand, when I export the same ADM mix to other object-based audio production tools like Orphus ircam admix renderer it doesn't even play it. So, what's the use of creating an object based 25 minutes film audio mix using EAR production suite when at encoding stage it doesn't properly encode ADM where other software's just simply rejecting it.... Your software is able to open other adm mixes from other production tools like Dolby Atmos production suite for example where we can clearly read and write automations to objects but when it comes to encoding or exporting stage your adm export might have some bug. Please check that reaper adm extension once again. Alas when there is no way that I can encode my immersive audio mix to Dolby atoms simple because of this adm recognizing issue, the chain is stopped. now I can't send my mix to any of our project reviewers. Alas I have to convert my object based to channel based 5.1. No use of preparing an object-based mix using EAR production suite when it ultimately lands to a channel-based mix that too less than 10 channels because neither windows or mac os support or recognize more than 10 channel wav files. That means if I create and record a channel based 3d mix like 7.1.4, a 10-channel wav file using spat revolution, it's no use because for channel-based content can only support up to 7.1 wav in windows and mac not yet 3d for now. I tried; we can't play at least 12-channel wav file on any consumer media playback systems to experience 3d. a 24 channel NHK 22.2 channel based immersive audio mix will definitely won't play unless until you encode such wav files to Dolby Atmos etc. For only you can deliver 5.1 successfully if you think of channel based. Hence ADM is very important and the only way if you want to convey your object-based audio mix not even ambisonics 7th order..... I am very disappointed to find that not working. Please fix this. Okay if this a problem with Dolby Atmos production suite or Admix renderer from ircam as they are unable to recognize your adm then tell me a way such that i should take an Adm mix out of EAR production suite and encode it such that my clients should listen my hardly worked and created 38 channel object-based audio mix on their 7.1.4 speaker system or any pair of headphones! as everybody says.... . There was a serious hard work involved and i did with musical perspective that is aimed at user's natural and realistic immersion into soundscapes rather than just playing with objects. Alas I have to end up with 5.1 . So please look into this issue and its compatibility with DAPS ....

junh1024 commented 3 years ago

Reproducing my own post,

"ADM was supposed to be a universal standard. However, the current situation is far from ideal. Different tools support different subsets, creating incompatibility. There are at least 2 "types":

As a workaround can you render 514. That's directly accepted by the Dolby Encoder for EC3 Atmos.

mpsrikar commented 3 years ago

What? what workaround? So, you are telling me to make a 5.1.4 channel print from EBU reaper and upload that 10 channels wav file to a dolby encoder? Or other? Please explain clearly to render this of my object-based audio to DAPS. I don't want to convert it to a channel based even if it's a 3d channel print. Because translating it to a dolby atmos equipped cinema theatre means that my mix will be decoded to an array of speakers and no more behaves like pin point object and the same old story dates back of sweet spot and resolution and phase problems of , 5.1, 7.1 systems etc.

junh1024 commented 3 years ago

The 514 workflow I described is EC3 Atmos for Home theater delivery.

If you are delivering for cinema, you can try exporting to 714, 914, (or other CBA), and then panning those speakers as static objects in a Dolby- compatible software, eg, Pro Tools or Resolve. Then export DAMP/ADM for DAPS.

It is a shame, but CBA is probably the only reliable way to get around OBA/ADM incompatibility at this time, and there are anecdotes that suggest people are using the CBA paradigm for multi-format delivery (eg, mixing in Atmos & exporting CBA for Auro-3D & DTS-X cinemas).

Disclaimer: I don't deliver for cinema so YMMV.

mpsrikar commented 3 years ago

okay here is my demo object-based audio mix. I remixed this in immersive object-based audio using the stems provided on a VIRPP remix contest platform. for ease of monitoring, i am giving you this immersive stereo format. i.e. This is in 3D BINAURAL format so use your headphones for evaluation.

This of my example remix sounds punchier and crispier due to the usage of nearfield HRTF synthesis. The most important and interesting part is this time i had fun with audio objects bouncing stuffs all around. There are many sounds in this song that i have dynamically moved which were originally placed static by the author. The user entertainment takes are as follows ..... Song contains 2 different synths originally out of which 1 synth i made it so ambient as it almost sounds as granular synthesis and ASMR type of wavy texture that is more appreciated towards end of song but originally it was missing but still there by authors. The second synth which was originally panned hard left by author that starts somewhere in middle in compliment to a trombone but is contains a frequency sweep. So now i took the advantage of that frequency sweep to further enhance it while panning it upwards. The synth now starts at left back and slowly moves upwards shooting directly above head during its sweep and then suddenly falling off again at left back and this process cycles. Binaurally we hear sounds from above with high frequencies hence i got the idea and tool the advantage of frequency sweep to pan things from bottom to top in left back and the results are literally so epic as real making the effect obvious which is very entertaining and fun user experience ultimately. Secondly there are Orgon sounds that feels like being pushed and swept away from right to left top and some funny fx sounds that are auto panned with lush reverb for lazy sounds like trombone's etc. that are left statically positioned right back as author intended. QA whether you are able to appreciate the ADSR and that cool panning .

Now i need to deliver this to a CINEMA. I am having all the object metadata as for now somehow need to get this working . EBU helped me a lot in preparing the initial mix for these audio objects using stems from there i transcoded it to binaural

Google drive link is attached

https://drive.google.com/file/d/1mji5pp9YlYC7nmfFrDLRlCXblpg-4UDO/view?usp=sharing

junh1024 commented 3 years ago

The song itself is mediocre. But, you've managed to make the experience exciting with creative panning. With the 20BIN mix, I can hear something is happening, mainly L<>R motion, but that's about as far as I can hear due to binaural not working well for me, so it's good that you described the action.

Anyway, Atmos OBA is will not totally prevent the issues you describe (like, sweet spot aka stability issues). It also depends on a few factors:

1- Your mixing. Try to pan sounds around the edges to avoid stability issues. No OBA system will save you if you did bad panning. I can't check this in your mix as you delivered 20BIN rather than 714 or other CBA (I can analyse & downmix if needed). You can encode 714 ogg vorbis q=0.4 if desired. 2- The cinema. Cinemas vary in speaker layout and might have less speakers than you thought. If you're delivering for a specific cinema, choice of CBA layout might be more relevant. 3- Dimensional paradigm. I think it's better to use 3D CBA rather than 51/71 since sounds in the absolute middle will be somewhat more stable at the top than at the middle. Also, people can hear your hard work in 3D. If you did some sounds @ back & height, The average audience member will prolly be happy. They may struggle to hear the difference between a sufficiently high-order surround & "true" OBA.

You're lucky in that you have access to DAPS & a cinema. I don't.

You may have a deadline to meet, and you want results now. Just use a CBA export for DAPS. I'm not sure what sort of magic solution you were expecting since I'm not a developer of this repo & my knowledge of ADM is limited.

Would you like a 2nd opinion from someone else?

PS: what speaker layout do you monitor on, with REAPER?

mpsrikar commented 3 years ago

Thanks a lot for your feedback. It's a regular 5.1 ITU SMPTE. However, since you asked, I do have a demo of my mix how it sounds on a typical 5.1 arrangement with regards to panning towards edges. You can QA it . By the way my past mix, you need not worry as you might almost felt it correctly despite the fact bin has personal hrtf issues because knowing this fact, that is why i intentionally used nearfield binaural synthesis which has strong spatial presentation without loss in timbre and clarity as head shadowing effect is less in a nearfield synthesis. Usually this is recorded by simulating near objects less than 0.2m around a dummy head. This technology has been adopting lately as a next gen solution for binaural synthesis by many tools like Dolby Atmos Production Suite Bin setting " NEAR " inside renderer, Flux immersive Spat Revolution (recent in budding stage) and Dear Vr pro (only when disabled direct and reverb signals and panning with early reflection but only when bringing object very close to the center ) etc. . This will now also allow users to simulate a bee flying around a head razor sharp with no compromise in spatial resolution vertically above head, 90 degree left and 90 degree right while it will not color and diminish the bee sound stem at all. As a summary L<>R and vertically above are only 3 zones in a nearfield synthesis that people will likely hear spatially and these zones are in more resolution that a standard HRTF binaural where there is a cone of confusion for most people. For example, using HRTF default at 2m and audio object panned straight front top might sound exactly same for some while it can sound at back of their head to others but again these far field HRTFs are important especially when audio objects are anywhere in between except 3 zones i mentioned above. For example, if you want to simulate a thunderstorm which is omnipresent in real world and is a 1 km away u can use default hrtf as it can deliver great sense of surround sound all around you but if you want your object addressed strictly to a specific location that too very near then you can opt for nearfield. In simple words to put on objects like slap delays and reverb pan then using far field hrtf and for percussive elements in the mix just pan them in those 3 zones (tetrahedral space) for better results. Do not worry about using and mixing two different HRTFs its no sin and at the end its all about perception of sound. Nearfield synthesis is also backwards compatible i.e if someone accidentally plays this on a stereo loudspeaker setup it will not sound that bad ONLY things panned directly above head will be a loss (prone to Cross Talk) at such situations. ALL in all Nearfield is best in my opinion to make sounds very sharp at left, right and directly above the head than traditional HRTF while also making it compatible for normal stereo. All these results are based on my personal research in this area.

Google drive link is attached to access my composition in 5.1 while a SoundCloud link is also provided to evaluate it in 3D Binaural again to understand how a loudspeaker render translates to Binaural and like A/B testing. Here are those links

https://drive.google.com/file/d/1ePaawEp-TDl4uFvmKXnsGZ47s5mhP3Z_/view?usp=sharing