#63

Thanks a lot, I’ll pass it along!

1 Like
#64

I haven’t done anything extremely experimental with the ECR-1. I’ve only been using it as an end of chain effect, but taking advantage of the CV modulation. I’m happy with the results from it, but I would recommend waiting on getting one until they release a new firmware for it. The current firmware doesn’t allow for saving/loading presets. You can save impulse responses but if you restart your system you will have to select your impulse response(s) and adjust all your settings again. If you’re someone who leaves their system on all the time or start fresh every time maybe this won’t be an issue for you. The firmware is hardware based (Raspberry Pi Compute Module (I think)) so if you do get a unit now you would still have to buy a new firmware from them whenever they release it.

#65

you will absolutely want the smearing effect of an allpass filter (or three) and to tune it by ear - if you don’t have easy access to a specialised one (plugin), try the shortest reverb you can find.

#66

I’ve never used it but the process they go through to capture multiple IRs in concert halls for use in the VSL MIR plugin is impressive! For example the speakers they use for replaying sweeps are on stepper motors, so they can ‘direct’ the source etc… More details here (which inevitably makes me wonder how this could be achieved outdoors!?)

#67

Last night I was looking at the Tasty Chips GR-1 and ECR-1 and it made me wonder what would happen if you gave a convolver a granular playback engine for the impulse responses instead of playing them linearly. Would that even be possible? It would be very cool if you had real time control over the granularization of the IR. I played with the idea a bit last night, running IR samples through a granular plugin and loading the result into a convolver, but without control over the granularization as it’s happening, it’s not much different than just loading up any atypical sample in the thing.

1 Like
#68

Could try this with the live convolution mode of the Mungo c1. Were you more interested in the “scattered grains” effect or in time stretching / pitch?

Wasn’t there some one-off custom hardware granular convolver? I wonder what its workflow was.

1 Like
#69

I think the scattering grains effect would be most interesting. I’ll see if I can find anything about a hardware unit–I’d love to see what that is if it exists.

EDIT: Is this what you’re talking about? Looks like it was given to a group of people as part of a Red Bull Music Academy thing. Looks interesting but I haven’t made my way through much of the presentation yet.

#70

I guess you could have a granular oscillator and record short sections as single use IRs. A set of rolling buffers. I think possible, but likely quite cpu intensive as you would be effectively running many reverbs at once.

#71

Yep, that’s the one.

I can tell you based on my previous forays that scattered grains as an IR sounds almost exactly as you’d expect! Convolution is commutative, so you can kinda technically reverse the IR and the input and get the same result (only “kinda” because in the audio world we tend to have the kernel/IR as a “discrete” component versus the input audio as a “continuous” component and then listen to the results one sample at a time). Give it a shot: take your synth sample or whatever and cut it down to IR length, then convolve little grains of filtered noise or whatever with it (voices and drums are fun) on the input side. Very vocoder-ish. c1’s sample rate controls make this patch into a tuned resonator kind of setup (though I don’t know if it tracks V/oct).

The length of the kernel matters a lot! As does its enveloping/dynamics. In fact I really, really wish that the c1 had a mode toggle for applying a simple decay envelope to the kernel in the live convolution mode.

1 Like
#72

sure. convolution is just the multiplication of two power spectra. in realtime audio convolution engines, we don’t actually convolve/multiply with the entire IR, because the latency would be too high. instead use overlapped smaller FFT windows. effectively, we are multiplying the results of two STFTs; one of them is transforming the input, and the other is (pre-)transforming the IR in a loop.

that “in a loop” is the part where “time stretching” the IR would get a little weird to implement in realtime. i guess you would want to capture the output of the granulation to another buffer/delayline with a known length. you would be modulating the length. as @_mark points out, to really get the effect of having a longer / shorter IR, you would also need to modulate the amount of overlap in the IR STFT, so that at any given moment you are hearing the effects of convolution with each part of the IR - and this would have CPU impact, so the amount of stretching you can do would be limited.

this should be pretty easily to implement in max/pd/supercollider. the granular convolver (in which @glia also expressed an interest, around the time norns launched) was implemented in supercollider, and i believe also took advantage of the wide array of phase-vocodor processing ugens available there. (to e.g. perform pitch shifting before convolution.)

but yeah - expensive, and i’m honestly not sure if the sonic effects of granulating the IR would be all that different from granulating the result of convolution.

3 Likes
#73

Here’s one I made from a recording of my bath draining

Listening to it on its own, it could almost be used as a Donk too haha

2 Likes
#74

Shirobon just released an impulse response pack of the PlayStation 1 CD player reverbs! https://shirobon.bandcamp.com/album/ps1-reverb-impulse-responses

Something about the concept of burning a sweep to cdr and sampling the response this way appeals to me immensely

Edit: also this raises the prospect og mastering one’s tracks by burning them to CDr and re-recording from the Ps1 :slight_smile:

6 Likes