Vember Audio Surge is now open-source

[uOpt]

So how would I go about making my own IRs to contribute? I have a few amps and speakers that might be interesting, plus suitable microphones.

Set the speaker and mic up as if you were recording anything else. Then somehow playback either a sine sweep or a full-spectrum transient through the amp and record the output of that.

That's all? I thought there would be some complicated pulse collection that can be tracked down individually in the output.

I'll see what I can do. Thank you for the information.

[EvilDragon]

Well actually, after you record a thing like that, you'd have to deconvolve if against the test tone that you sent into the device, in order to get the pure impulse response. Voxengo Deconvolver is a pretty great tool for that, but it's not free. However it does have a demo mode which is functional to an extent (3 deconvolutions per session).

There's also this one, which is free: https://wavearts.com/products/plugins/ir-capture.

[uOpt]

Looks like a nice learning opportunity.

ETA: Logic seems to have a module for it, too. https://support.apple.com/guide/logicpr ... 58c7a4/mac
https://books.apple.com/us/book/impulse ... d960809866

[audiojunkie]

Well actually, after you record a thing like that, you'd have to deconvolve if against the test tone that you sent into the device, in order to get the pure impulse response. Voxengo Deconvolver is a pretty great tool for that, but it's not free. However it does have a demo mode which is functional to an extent (3 deconvolutions per session).

There's also this one, which is free: https://wavearts.com/products/plugins/ir-capture.

Can you capture other effects, or is it mainly for reverb and cabinets?

[Surgo]

Well actually, after you record a thing like that, you'd have to deconvolve if against the test tone that you sent into the device, in order to get the pure impulse response. Voxengo Deconvolver is a pretty great tool for that, but it's not free. However it does have a demo mode which is functional to an extent (3 deconvolutions per session).

There's also this one, which is free: https://wavearts.com/products/plugins/ir-capture.

Can you capture other effects, or is it mainly for reverb and cabinets?

It can capture anything that can be described as a finite linear filtering process. In practice, that means reverbs (including cabinets), speakers, and instrument bodies are the most common thing to be captured.

[EvilDragon]

Yeah, any pitch modulated or distorted things are not going to be captured well.

[planetearth]

You could also capture pre-amps, EQs, or virtually anything that changes the audio signal. The only issue is that you get an extremely small "snapshot" of the change to the signal. That's why some things work better than others as IRs.

You can find a few more tips here: https://thegearforum.com/threads/is-the ... nses.3820/ (And of course, your fellow KVRers.)

And Wave Arts has a free tool to create IRs: https://wavearts.com/products/plugins/ir-capture

Steve

[Andreya_Autumn]

That's all? I thought there would be some complicated pulse collection that can be tracked down individually in the output.

I'll see what I can do. Thank you for the information.

Yeah the deconvolution that ED mentioned is the complication I guess. Which is basically when you use a pre-recorded impulse it's possible to sort of clear the impulse itself out of the response in an automated fashion.

You do not always need to do that, and quite commonly people don't for reverbs. Recording acoustic spaces in particular some folks just pop a balloon and then maybe edit the pop itself out of the start of the file. One of the ones we will use is literally just a recording of someone smacking the soundboard of a piano with the sustain pedal down (recorded with really good gear). Works fine.

But yeah with amps/speakers in particular it's usally advantageous to use the deconvolution strategy. It may be impossible to get a clean result without it in fact. Hence the suggestion to send us the impulse file so we can do that. But you can also do it yourself indeed!

[uOpt]

So in first summary, no making new overtones via clipping of any kind or direct pitch shifting.

Thanks, folks. Very educational.

[Andreya_Autumn]

That's not quite what ED meant. It's more like "if the process you're taking a response of is nonlinear (i.e. distorting or otherwise level-dependent) or has modulation that evolves over time (of pitch or otherwise), an impulse response can't fully represent the sound of it."

Which makes sense when you think about it. An actual speaker would colour the sound slightly differently at different volumes, but the IR will colour all sound as if it was played at the level the IR was recorded at. A modulated reverb will have pitch modulation, but an IR of it will sound as though the modulation is "frozen" at the position it was in the moment you took the IR.

That doesn't mean don't take IRs of those things! The ones I did the other day have those properties, spring reverb is really nonlinear and tape delay has flutter etc, but they still made for interesting sounding IRs. And speaker IRs are obviously super popular even though they're theoretically "flawed" if you will.

[audiojunkie]

That's not quite what ED meant. It's more like "if the process you're taking a response of is nonlinear (i.e. distorting or otherwise level-dependent) or has modulation that evolves over time (of pitch or otherwise), an impulse response can't fully represent the sound of it."

Which makes sense when you think about it. An actual speaker would colour the sound slightly differently at different volumes, but the IR will colour all sound as if it was played at the level the IR was recorded at. A modulated reverb will have pitch modulation, but an IR of it will sound as though the modulation is "frozen" at the position it was in the moment you took the IR.

That doesn't mean don't take IRs of those things! The ones I did the other day have those properties, spring reverb is really nonlinear and tape delay has flutter etc, but they still made for interesting sounding IRs. And speaker IRs are obviously super popular even though they're theoretically "flawed" if you will.

Ah! This is beginning to make sense! So, the capture is a sort of snapshot of a moment in time, and there is no modulation in a snapshot. This is like a single cycle waveform vs sweeping through a wavetable. So anything that can be captured as a snapshot would work, but anything that would require sweeping through multiple snapshots won't work--like modulation over time.

Am I getting closer to understanding this?

[Andreya_Autumn]

Yeah for sure. "Won't work" is a bit strong but replace with "might or might not sound satisfyingly true-to-life" and you're pretty much there.

[Jbravo]

Can anyone tell me where the speech wavetables are stored, as in when you use TWIST engine and vowels/speech. There are a few phrases there, but is there a way to add more or edit those? Can I do the same with just a regular sample and wavetable setup?

[EvilDragon]

They are hardcoded in C++, they are not stored on disk.

[Jbravo]

Ok, thanks for the info