Making your own impulse responses

[manisnotabird]

I've been experimenting with sending a very short burst of noise from my Minibrute through a hardware reverb to make my own impulses for the new RV7000. Any other/better ways to do this on my Mac without buying expensive software?

[manisnotabird]

Having fun experimenting with recording variously filter and filter sweeped noise from my Minibrute into the convolution part of the RV7000.

[JoshuaPhilgarlic]

I've been experimenting with sending a very short burst of noise from my Minibrute through a hardware reverb to make my own impulses for the new RV7000. Any other/better ways to do this on my Mac without buying expensive software?

It's said that the perfect signal would be just 1 sample. I tried to create one in Protools, but I can't say if it's perfect, 'cause it's too small even for Protools .

It's 192kHz/24Bit

I checked it with TSAR-1R, and with a little tweaking of send level and RV7000 Decay it sounds pretty close !

[ScuzzyEye]

It might be slightly better to use a wave file that's at the sampling rate of your Reason project. This will keep Reason from low-passing the file, and resampling it to the current rate.

I've made a collection of impulses that are at all the sampling rates that Reason supports. I wrote a program to generate these (might make it into a free, utility RE). They are 32-bit floating point waves, and definitely contain one sample at 0 dBfs, and then a second of silence after that (just so you can find the audio in your track).

I'll attach it to this post, and it's also linked here: http://scuzzyeye.com/patches/impulses.zip

[vectro]

I use Impulse Response Utility which I got with Logic, on my Mac.

On Windows, use Voxengo Deconvolver to do it.
http://www.voxengo.com/product/deconvolver/

Or else, use a really short burst of white noise (yep, all frequencies are needed equally to not affect the tone to become brighter or darker). A start pistol is popular when doing impulses in real acoustic environments.

[manisnotabird]

I use Impulse Response Utility which I got with Logic, on my Mac.

That's what I meant by "expensive software". :/

[ScuzzyEye]

Or else, use a really short burst of white noise (yep, all frequencies are needed equally to not affect the tone to become brighter or darker). A start pistol is popular when doing impulses in real acoustic environments.

When dealing with analog, yes, you need all frequencies. But because of how digital works, a single sample at max amplitude, preceded, and followed by an infinite amount of silence (well, enough that the device you're measuring is silent) will produce the impulse response (by definition).

The reason to use an impulse that's at the same frequency rate as your Reason project is that, as I said, Reason will low-pass filter a sample of a higher rate before playback. That effectively adds the impulse response of that filter to the impulse. Granted the resampling filter in Reason is pretty good now, so it'll have a minimal impact, but avoiding any possible side-effects is always the best way to go.

[Julibee]

It might be slightly better to use a wave file that's at the sampling rate of your Reason project. This will keep Reason from low-passing the file, and resampling it to the current rate.

I've made a collection of impulses that are at all the sampling rates that Reason supports. I wrote a program to generate these (might make it into a free, utility RE). They are 32-bit floating point waves, and definitely contain one sample at 0 dBfs, and then a second of silence after that (just so you can find the audio in your track).

I'll attach it to this post, and it's also linked here: http://scuzzyeye.com/patches/impulses.zip

Thanks, Scuzzy.  I'm having way too much fun with this stuff... 

[vectro]

a single sample at max amplitude, preceded, and followed by an infinite amount of silence

True, I'd do that whenever making impulses of signal chains, either in the box or through hardware.
But when dealing with capturing acoustic environments I still prefer sweeps or noise bursts.

[manisnotabird]

When dealing with analog, yes, you need all frequencies. But because of how digital works, a single sample at max amplitude, preceded, and followed by an infinite amount of silence (well, enough that the device you're measuring is silent) will produce the impulse response (by definition).

Does it matter that the signal is converted to to analog, then into digital, then back into analog, then back into digital in the signal chain I'm using to "sample" my hardware unit (right now, actually a cheap Digitech multi-efx guitar pedal).

[avasopht]

The conversions between analogue and digital will be imprinted into the impulse response.

If you bypass the effect you could get the impulse response of the filter chain and deconvolve the reverb impulse response with it to remove the effect of the filters.


---

Regarding noise, .. the longer the period the more the phase is smeared towards 0, affecting the higher frequencies first, killing any interesting unisons that might be happening.

[manisnotabird]

I was inquiring whether the conversions mattered with regards to the choice between using a single sample "click" or a short burst of white noise.

[ScuzzyEye]

I was inquiring whether the conversions mattered with regards to the choice between using a single sample "click" or a short burst of white noise.

A single sample impulse, should be rendered in the analog domain as a very short burst of all frequencies below Nyquist. But that's not going to happen perfectly. So what you end up with is an imprint of the DAC on the impulse.

If you send the single sample impulse out through your the DAC/ACD chain without the pedal (or even better the pedal on pass-through) you'll get the impulse response of the digital to analog to digital conversion. There are tools that can remove that signature from the IR that you get with the effect enabled. That'll be a better representation of the pedal alone, but it might not be so bad if you don't go to the extra work. Try it and see what you think.

[vectro]

Greater audio interface, greater results.

[thala]

Greater audio interface, greater results.

if you try to rebuild something exactly; you are right.
i found my dream impulse yesterday:
i took driver with its noise osc and some colour modulation. then i recorded long impulses up to 4bars at 120bpm.
these noises i used as ir. already nice. but then i used the size knob and set it to maximum 24. i think these are semitones. so two octave down. sampled at 44khz. which result in 11khz effectively in use. and what there came out... deep verb is obsolete now. a damn beautiful sound, that i was searching for years. thanks to props. and thanks to ochen. this guy is just happy.
the best thing: if go extreme with long noises, the decay knob scales it very well down
so there seem to be no need for short ir. the decay knob will fit the size very well.

[Gaja]

Hey thala, care to share the patch (probably best in the IR thread sticky)?

[thala]

Hey thala, care to share the patch (probably best in the IR thread sticky)?

here it is.
and a soundexample with endless spaces on a abl3-preset. if you put echobode on it, you will get a another synth sound

i think its best to post it here. if linus want this and/or more he can pick it up. over there are the "pro" IRs, that try to sound realistic.
maybe there should be another separated thread for more selfmade/experimental/freak stuff?

[Gaja]

Thanks for sharing! I'll check it out when I'm at work.

[Ostermilk]

Here you go, a set of Dirac Pulses in any form you want in wav format.

http://www.noisetime.com/impulsecreation.html

These are the real deal full spectrum pulses so you won't get any tonal bias like you might with a clapper board, gunshot, baloon pop or other form of impulse.  Play the Impulse through the device or into the space you want to capture and record the Response.  Trim of any space (and the impulse if preferred) you don't want at the begining and the end et voila one Impulse Response to load into your reverb.

So if you can't deconvolve a wet swept sine these are the way to go.

A good way of using them for ITB stuff in Reason is to put one on a bar line of an audio track and include any effects you want to capture set that track as a 'Rec Source'.  Create another audio track and set the inputs to the impulses track then hit record, make sure you leave enough on the end to capture the full respone than simply bounce this recording to disk.

Remember you won't be able to capture any time variant effects such as saturation, chorus etc, but you may well end up with something unexpectedly interesting by experimenting.

[ScuzzyEye]

When using a very special impulse called Dirac impulse (a single maximum spike)

So the same thing that I made. Yeah, there are more formats, but they aren't needed for Reason, and no 192kHz. Maybe the stereo versions might save some a step of hooking up an extra wire for devices that don't handle mono inputs the same as they do stereo.

[Ostermilk]

When using a very special impulse called Dirac impulse (a single maximum spike)

So the same thing that I made. Yeah, there are more formats, but they aren't needed for Reason, and no 192kHz. Maybe the stereo versions might save some a step of hooking up an extra wire for devices that don't handle mono inputs the same as they do stereo.

No need to be precious, I simply hadn't seen your link previously and posted in response to the OP who was asking for the best method of creating responses without resorting to expensive software.

Also for anyone else interested there's a free deconvolver available for Windows only though which is an alternative to Voxengo's excellent (and inexpensive) commercial deconvolver.  It's called 'Gratisvolver' and it will allow the use of the swept sine method.

http://www.catt.se/download_area.htm#Gratis

I'm not aware of anything similar for the Mac platform but other Mac users may chime in with something and I'm also sure if you use something like Matlab there are plenty of snippets of code lying around that will do the same job.

Disclaimer: *Please note that this information is provided in the hope it might be helpful to anyone interested.  In no way is it intended to diminish anybody's particularly ScuzzyEye's already stellar work in this field which I hereby acknowledge and I'd like to take this opportunity to apologize to him profusely for missing his previous post.*...

[ScuzzyEye]

I do sincerely apologize. I had been up for about 20 hours at that point (trying to get my RE finished). Reading what I had written does comes off pretty poorly.

[Ostermilk]

I do sincerely apologize. I had been up for about 20 hours at that point (trying to get my RE finished). Reading what I had written does comes off pretty poorly.

NP.  I ought to have read the thread properly in the first place.

[selig]

Greater audio interface, greater results.

if you try to rebuild something exactly; you are right. i found my dream impulse yesterday: i took driver with its noise osc and some colour modulation. then i recorded long impulses up to 4bars at 120bpm. these noises i used as ir. already nice. but then i used the size knob and set it to maximum 24. i think these are semitones. so two octave down. sampled at 44khz. which result in 11khz effectively in use. and what there came out... deep verb is obsolete now. a damn beautiful sound, that i was searching for years. thanks to props. and thanks to ochen. this guy is just happy. the best thing: if go extreme with long noises, the decay knob scales it very well down so there seem to be no need for short ir. the decay knob will fit the size very well.

That's what I've been doing (more on this as I get time), but 24 semitone is two octaves, which means an effective response of 5.5 kHz (not 11 kHz, which would be one octave down or 12 semitones). 

[Ostermilk]

Some more food for thought on rolling your own can be found in these excellent ol' articles from SOS.

https://www.soundonsound.com/sos/sep10/ ... lution.htm

https://www.soundonsound.com/sos/sep10/ ... matlab.htm