spatial audio in Bitwig

[ThomasHelzle]

I've been able to have a play today. Thanks again for sharing @ThomasHelzle - it's been very helpful to try out. It took a bit to figure out how to do the binaural mixdown (I'm still using Waves, sadly but was able to get it to work with MetaPlugin).

I've noticed the panner tools (LFO, Random, Step) all seem to freak out for a moment when adjusting the rate. Is there a way around this? I'd like to be able to ramp the rate live during performance.

For live work I've built a quad performance mixer in VCV. It supports multichannel FX sends along with a few other modular niceties. Eventually I'd love to port it into the grid, but that won't be possible until Bitwig support multichannel.

Yeah, the binaural part is definitely not straight forward...
Both Waves NX and IRCAM hEAR are a bit wonky, I also tried DearVR but none of them does convincing back/forth or up/down. Still I somehow like hEAR best on the headphones, although their GUI is broken badly - and Waves is just a totally f**ked up company I don't want to have any part in anymore.

You can change the LFOs etc. to not use the synced mode but set them to free running...

I had also built a panner in Reaktor, back when Bitwig was more limited, but it was dragging down performance with many instances: https://www.screendream.de/code/toms-surround-panner
How is VCV/Cardinal in that regard in your experience? I would assume that having it on every track could be a performance issue as well?
The Bitwig panner also has a penalty to it of course, so be sure to disable all the sends you don't use, that helps a bit.

Cheers,

Tom

[hockinsk3]

Many spatial audio guys generally don't use binaural because no HRTF translates the source signal well, it's fine for gaming, podcasts, low energy music like acoustic and single room recordings etc but you always experience the weird phasing issues, filtering etc with binaural filters and hrtf hrirs etc.
In my opinion, it's superior to implement your own localisation approach using amplitude panning and reverb techniques. This is why Mach1 don't provide any HRTF trickery because their customers belief is you shouldn't be using it if you want total transparency and translation into other spatial delivery formats when monitoring.
The most transparent binaural I'm aware of other than the Smyth A16's is actually a new non-HRTF approach from Kinicho which I'm beta testing at the moment. REally quite impressed, it resolves many of the issues with HRTF.

[hockinsk3]

...

[hockinsk3]

I requested it quite often, but it surely would help if you guys send request for multichannel tracks as well to the support please.

I'm hopeful this is coming, especially with the move to 3rd party development devices. Discussing it with Bitwig (i remind them every few weeks mutlichannel audio receiver and ability to set/change channel format of a VST input chain would resolve so many workarounds we're currently having to implement) they do at least say multichannel isn't ruled out. Under the hood, Bitwig does already support it from a technical ability, it's just got to be coded. When you dig deep into some of Bitwigs code, there is even a channel count for tracks currently set to 2.0 so that gives more hope too!

[ThomasHelzle]

Interesting - thanks hockinsk3!

I don't invest any money into this stuff anymore. My panner works for my personal needs and the binaural thing is more for having a more natural headphone experience than real spatial listening.

I liked the Ambisonic Toolkit (ATK) for Reaper when I tried it, especially the super-stereo-encoder did some really magic stuff from stereo sources, like Laurie Andersons "Monkey Paw" (one of my go-to audio test songs), so when Bitwig should go multichannel, I'll have another look.

And yeah, the whole group-structure basically shows a kind of multichannel idea already, so I can imagine it being somehow in there like you say.

That latest move with the paid addons could also mean we will look at some kind of Bitwig Suite in the future, which could have advanced features like multichannel, ambisonic, video support etc. maybe. No idea where they are going though, so it's all idle speculation at this point...
Also not sure if I would be willing to pay even more than the current semi-subscription.

Well, we'll find out sooner or later

Cheers,

Tom

[hockinsk3]

You're wise not to invest much, there's nothing commercial out there that's really any better than what is free or buildable yourself in terms of working in a DAW imo. You can make a superior Binaural HRTF than Dolby Atmos provides with their Renderer for example using free tools and Sparta for example. I bought Mach1 as much to support them as a small company with their SDK & Transcoder development, than me needing their panners, because they are only amplitude vector panners, you could just use Bitwig Vector modulator I guess (multichannel outputs are still difficult though so generally you end up in the Grid which is both good and bad). Their Mach1 Monitor is a little proprietary to them, but it's not changing the sonics at all. I think this is why I like it, because you can mix spatially and then transcode to the target format you want, be it Atmos or Ambisonics etc, but it still sounds like the stereo/mono source and so you have 100% control still.
I'm aware GPU Audio guys are working on something with Mach 1. Can't say much about it, but GPU Audio will bring the ability to process perhaps 100's of channels of audio in parallel with almost zero latency without the concept of multichannel tracks and busses. ie it will work in any Stereo DAW (Fiedler Spacelab also works like this in stereo DAW, but is CPU-based audio so a bit DSP heavy).

[ThomasHelzle]

Thanks again, very interesting!

I tried the GPU audio stuff they have in beta and wasn't impressed one bit.
If they can pull off their claims, great, but their first demo plugin was totally lame and needed more CPU on my machine than the Bitwig inbuilt convolver, so as far as I'm concerned, they haven't shown yet in reality(TM) that they can deliver and it would need a whole eco-system of GPU based DAW and plugins to make it _really_ fly AFAICS.
I left the beta since it just didn't seem to make any sense, they defaulting to 96kHz and super low latency felt also silly, since that isn't something the average user can use in a realworld production.

So yeah, the future might be bright for multichannel, but it isn't here yet

Cheers,

Tom

[hockinsk3]

GPU Audio wasn't even Beta it was pre-beta early access just to get some real-world testing on multiple DAWs and hardware to see where the issues were. Unfortunately, Bitwig has a known compatibility issue due to how the VST callback is a little different compared to other DAWs iirc. (I think GPU explain this on the Discord). I've been impressed with it outside Bitwig, easily running 100+ Convolution instances without latency is worth pursuing for them imo. Early days though.

[trevormeier]

I've largely switched to creating Dolby Atmos in something called Mach1 which is far more transparent and simple way to get to Dolby Atmos from Bitwig.

Ah, got it. I'm not in a place where I can buy another plugin suite right now so I'll keep experimenting within Bitwig. Thanks for the pointers though!

[trevormeier]

You can change the LFOs etc. to not use the synced mode but set them to free running...

I discovered this after further experimentation last night. Setting them to note sync could allow me to use a note-on event to sync things if I want, while otherwise being free-running.

I had also built a panner in Reaktor, back when Bitwig was more limited, but it was dragging down performance with many instances: https://www.screendream.de/code/toms-surround-panner

"Tom's Surround Panner" was actually the very beginning of my journey into Bitwig! I was searching for solutions to my problems while working in Logic + VCV and ran across your plugin, which led me down the long and winding trail to here...

How is VCV/Cardinal in that regard in your experience? I would assume that having it on every track could be a performance issue as well?

The Bitwig panner also has a penalty to it of course, so be sure to disable all the sends you don't use, that helps a bit.

I did a few quick tests. On 32 tracks, Cardinal/VCV uses about 350% CPU (about 11% per track). Your surround panner just sending to front and rear uses about 280% CPU (9% per track). I built a quick quad panner in the Grid, sending out via BlackHole and having front/rear FX returns monitor the same channels as input, and it uses about 180% CPU (6% per track). My little quad panner doesn't have any modulation, so I'd say they're all within a similar zone performance-wise. The Cardinal panner has a bunch of extra things going on and no work to make it efficient as of yet, so I suspect it could be much lower with a little work. One advantage of the Bitwig-native panner though is that CPU usage drops significantly when transport is stopped. Cardinal stays about the same.

It's really too bad that the FX grid isn't able to add chains or otherwise send more than stereo out without resorting to software loopback. For such an elegant piece of software, it's a pretty glaring omission IMO. Even before adding true multi-channel support, just allowing additional chain outputs from the grid would enable so many workflows.

[ThomasHelzle]

Ha - great to hear that

And very interesting findings.

Yeah, somehow the Grid is a bit baffling in some aspects. Since initially Bitwig intended to open up their under-the-hood modular system, I was confident that when that would happen, all my problems would be much easier to solve. But while I understand basically why they didn't go down the Reaktor route but did a much more approachable thing with the Grid, I'm still a bit sad.
I'm about to move away from NI as much as possible, so a Reaktor replacement, but hopefully better, would have been nice.

Well, ATM I'm more into realtime graphics with tooll3 (https://github.com/still-scene/t3), cables (https://cables.gl) and looking into some other systems that also support real surround capabilities, maybe one day I'll find something that is a bit more open than Bitwig even...

Cheers!

Tom

[trevormeier]

Ooohhh that looks like a fun rabbit hole!

[Tj Shredder]

Looks a bit like Max/MSP/Jitter…

[ThomasHelzle]

Yeah, I do 3D for 25+ years and got a bit bored by the slow interaction in "normal" 3D software, so realtime is fun to explore...

I was looking at all the available node based tools for years and never found any of them appealing, vvvv is too node-heavy for my taste (I don't like atomic nodes), TouchDesigner has an ancient feeling GUI and does too much "behind the scene" stuff for a node based tool IMO, so the best thing about node interfaces - you look at the graph and get a basic idea what it's doing - is partially getting lost (which is sad, since it's origin is the same as SideFX Houdini, which is my main 3D app...) and Max/MSP never clicked with me either. Notch looks fantastic but is very expensive, some others are just for the Mac or subscription only or only run on heavy GPUs, while we currently try to save energy working on our laptops...

Tooll3 is very well laid out and cleverly designed, even though it's early days for this re-written version and it's lacking a ton of features the others have. But the good thing is, that I can grow with it, so I already wrote my first HLSL shaders and can contribute to the development and help with ideas.

Cables is fantastic since it runs in the browser and it's a bit more evolved in some areas.
https://www.screendream.de/motion/inspiration

Both aren't geared towards audio much, other than supporting a stereo soundtrack, but are good to learn the basics about realtime and both support Midi and OSC input. And both are made in Berlin, so I can meet the devs and take part in their user meetings.

The best for audio would probably be game engines, Unreals audio system is said to be amazing, but those are quite overwhelming for a first contact and less geared towards graphics.
Our interest is mostly stage design (something I did in the past before 3D), so game engines first need to be dragged away from their usual targets. Godot could be interesting and supports surround sound, but I didn't look into it deeper.
Epics interest in the CLAP format could become a major incentive in the future, but for now, those smaller toolls are great for where we are.

Fun!

https://twitter.com/SuKoMotion/status/1 ... 2901693441

Cheers,

Tom

[Cyoon]

To get back on topic

@hockinsk3 did you check out L-ISA Studio? I'd like to give it a go, but their pricing model is holding me back, seems it's really a subscription like, you can use it only as long as you pay. But I especially think that it's very helpfull to have a full view GUI on where the objects are placed, so considering it and it would allow for an easy Bitwig workflow with direct outs. Final mix down then in Cubase maybe.

https://estore.l-acoustics.com/en/softw ... -year.html

Interesting article here from Jean Michael Jarre: https://www.heise.de/news/Oxymore-Jean- ... 84839.html

German though^^

Mach1 doesn't seem to have 64bit plugins, does it? Couldn't find the information and when I installed the demo, in Cubase it throwed an error. So, not so nice to invest in 32bit plugins these days.