And are you on Github maybe?

I'm lately socialising more on Github than other social media, except for this forum and some Youtube.

 

Actually I plan on just designing the 3D models and rendering the impulse responses, and license this to an established plugin dev. While I'm familiar with the maths and DSP, I don't know anything about plugin development. I'm more of an acoustician, not a dev.

 

you are probably aware that position vs filter depends on the output master, i.e. the user normally needed to create a headphone mix and a stereo mix with each different aural filters.

one solution is to not include the "direction" part into the IRs or FIRs but instead use some kind of intermediate immersive format - followed by possibly different filtersets. how is your approach on that?

 

I don't plan on doing things like custom binaural rendering. I'm familiar with how to do this, it's just that I'm focusing on a more traditional style of reverb, recreating the sound of a room in a recording.

99%+ of people don't have the required setup to accurately reproduce the sound-field of a room reverb, which is either a HOA/WFS speaker setup or a calibrated binaural system with a custom SOFA file. There's no way for stereo speakers or standard headphones to accurately reproduce the direction of each reflections, which is an essential part of the experience. So I don't think it's worth it to do a "virtual monitoring" grade room simulation, like what is used in acoustical research. This reverb is targeted for artists and sound engineers.

However, one of the virtual microphone positions that I chose for rendering these IRs is an AB pair with omni mics seperated by 19cm, like average human ears. This allows for a realistic soundstage on headphones/IEMs (especially considering that they should be calibrated to a diffuse-field target curve in any case) that accurately reproduces the real experience of late reflections, but still sounds great on speakers unlike what would happen with a binaural microphone.

Yep, what could be very interesting is Ambisonics, which would allow easy downmixing for Atmos, surround, binaural etc... However I'm simply focused on stereo right now, and if I find some devs interested in this, I will discuss potential immersive formats with them.

 

okay, so you are not aware of that (maybe?) requirement.

not ideal, but good enough.

as you say, you can use A/B or dummy head mics and have the directional filtering in a room response for an IR file already inluded - one reason why we use this tech. and you can do the same with ambi or vpab, where you simply interpolate between a handful of positions.

no need to support such output formats in any way, it will all just happen inside the reverb algo.

what some reverb designers have been doing is to include aural panning FIRs only for the frontal 60 degrees - that is an option in between these methods.

however, the main component is probably how you use it in a mix across different virtual sources.

 

?

Okay I think was a little confused by what you were saying previously. So yeah, my ray-tracing rendering takes care of "direction" for the early reflections. I have virtual mics with frequency dependent polarity filters. So it happens inside the algo, for now


Anyway, as a I saw a few responses mentioning Teldex, I'll look into it. I'm currently working on IRs of a famous concert hall and I'll share a demo in 2 weeks or so.

 

i was just thinking about one of the room situations our ears like the most: concert halls. there, the rear really does not matter, these rooms are basically "only stereo", isnt it?

so if it for "good sounding" and not for recreation of every possible space, that is fine.

 

What do you actually mean by developing?

If you're writing it in native C++, you do have the option of writing modules in Synthedit, people crap on it a lot, but then they'll use JUCE and be just fine with it. The modules in Synthedit are C++, plus there's other GUI stuff.

Even if you never actually release the Synthedit VST, you can still use it for marketing, i.e. designing a GUI so whoever you're selling it to can actually interact with it instead of just a bunch of code.

That's also why some people use Synthedit, they make sort of lesser featured plugins, stuff that's easy if you know "the mathz" and link to it on their site to get traffic, then, sometimes, they get snagged by other Devs.

 

The more I look at what you've posted here, the more it seems like custom Synthedit modules make sense, because this seems like two separate processes:

1: Intensification of a signal
2: Degradation of a signal

The first one, you'd want to be isolating fundamentals and overtones, then said overtones, obviously, have their own overtones (i.e. any "sound", technically, contains all other pitches, albeit in lesser percentages), which degrade into some kind of noise, you said Gaussian earlier, the specifics of which I'm unaware, but the general nature of which, from what I understand, is a sort of non-linear interaction.

Keep in mind the overtone series, Schoenberg explains this, there's a the fundamental, then the cycles which are repeated the most frequently, i.e. the most "consonant" sounds in a major chord, as well as those sounds which are the secondmost "consonant", i.e. the overtone which is repeated first IN SERIES, but that becomes secondmost, in terms of total repetitions, overtime.

What kind of mathematics are you actually using to approach this? It seems like you're looking to shift from Physics into Statistics . . . And those you'd have to interrelate with some kind of waveform analysis, since they're different.

You might actually have to have consonance and dissonance settings or something like that. . .