KupoBeat

test world updated again, getting ready to plug EQ and Compressor VST in, for the moment just broke this out to multiple busses.

Audio Pump now has 5 busses,

4 with vst and the 5th is a raw passthrough (useful for audiolink)

Kupo's AudioPump Loft by KupoBeat AudioPump takes AVPro streamed audio and plays it out standard Unity AudioSources in realtime‚ allowing for Unity audio effects to function in VRChat․ Creative use of external calls to native C＃ funct...

AudioPump has been updated with GPU Shader powered processing.

The test world has been updated with the latest build. A 6700k cpu passes performance tests.

A FIR 8 band EQ and Wave Shaper/Limiter will be added in shader code, and in world controls, will be added soon.

vrchat.com/home/world/w...

Learned shader development. AudioPump is now running faster then ever.. stable stereo audio at 6fps with minimal udon sharp overhead. Previous AudioPump code needed 1-15ms per frame in purely udon sharp. Now to write an FIR EQ and Limiter/Wave Shaping compressor like filter in shader code.

Soon.

bits and bytes and floats oh my. I think I have a way to get data in and out of a shader at basically no performance overhead in AudioPump. This may actually be viable to do in capped instances.

Delaying releasing the full AudioPump for a bit- There's a faster way to implement an EQ and Compressor filter then in udon, going to be pushing audio into a gpu shader to do the hard work there. This should make it functional on even low end computers. Latency will be a little higher.

😅 ever wonder why streamed 5.1 content has muddy bass in vrchat avpro? It’s what OBS tells the aac encoder. OBS (and aac defaults) assume the subwoofer channel is meant for gunshots and other movie effects. Patch below is the fix for CoreAudio. DM me if you want the built result.

DJS IN FULL EFFECT. TONIGHT

This 2 week rabbit hole spiral is all @lightbulb.systems ‘s fault

This has been a fun side project and good introduction to Udon Sharp. Looking forward to creating new things soon.

A few vrchat venues are about to have audiopump basic powered reverb/lowpass/highpass ambience music in the spawn area to help test how the script performs at 80cap.

A future feature release could potentially to be to write an eq and compressor filter in shader code, and increase the latency timeline to allow for posting work to the gpu. This is a bit difficult in some sense due to not having a good forward state storage for things like compressor envelope.

The advanced usage mode can have features disabled on the fly, so a “potato mode” button could be put into a world. Not going to lie, turning every advanced feature on is probably capped to 32 players in an instance.

In the advanced usage mode AudioPump contains the following features (which will be quoted at 60fps/7950x3d frame time measurements)

Source volume tracking (Free)
Pre gain adjustment. (Free)
Post gain adjustments. (.3ms)
3 band parametric eq (3ms)
Hard knee compressor (1.5ms)

AudioPump has a few modes of operation.

In the most basic mode it takes audio from one place and outputs it at another place. In this mode it consumes very little cpu time. At 60fps on an 7950x3d it’s under .3ms per frame. At 6fps it’s around 3ms per frame. This “should” be fine in a busy instance

Debated releasing a public beta of AudioPump today but have been slammed with work and other project stuff. It’s still coming soon, wanting to get a finished logo for the project first too though.

Couple of notes about the project in a thread below-

Yeah nothing in my actual script is sync, by design, was trying to gpt some kind of external controller thing for this that does do syncing but it doesn’t work. I suppose I have to just learn how this all works the right way

it would be nice if it was, this is for the AudioPump test world

Kinda fumbling with UI stuff, backend is more my interest. How do you wire unity slider values up to control an udon sharp script?

I added pre and post gain control, and bypass toggles for the eq and compressor, which can work for a potato mode toggle.

Need to put a non interfering beta into some 80 capped instances to see how AudioPump holds up

This weekend maybe i'll build a synced mixer console toy.

each block process operation seems to hold up on "my machine", though i do need to see how it does on older hardware.

The compressor uses the same pre computed math and look tables with a generated unrolled loop function.

AudioPump works by using a Needle/Haystack search function to find the next index in GetOutputData(). Samples are collected into a stereo pair of float[1024] blocks for processing.

The EQ uses all pre computed math and lookup tables for performance. all functions are in a generated unrolled loop.

The latest beta of AudioPump has been uploaded to the test world. Added a basic compressor and 3-band eq. The compressor and eq controls are live update safe, so they can be mapped to an in game console. Reverb uses the native unity audio filter. Works at 6fps or higher.

vrchat.com/home/world/w...

70k something function calls for 8 band. need to get this down lol.

Udon Sharp be show lmao. This 8 Band runs in real time at 6fps in unity editor, but I have doubts it will work in a busy vrchat instance.

🤓