Impulse Responses suitable for ER-301

Given the release of the new Exact Convolution reverb for the ER-301 I thought it would be a good idea to start a thread for sharing ONLY IRs that perform well given the limited capabilities of this new unit.

By that I mean please don’t share anything that is going to exceed the capabilities of this unit, i.e. these should be curated IRs that are known to work.

Please do upload any you have created yourself - even if you think they are not professional standard or whatever. The beauty of these things is that crazy things can happen when you least expect it.

Please respect any copyright, but feel free to also share any links to resources that could provide food for this thread.

With a bit of community effort I think we can probably come up with a really nice selection that will really help us all make the most of this unit.



May i ask :

  • what are the specs for suitable IR ?
  • how do you step by step create one ?

EDIT : found this interesting article :
But still open for advice ! (especially, now that i got the principle : what impulses to use and what software to extract the IRs)

1 Like

IR files are just like other samples (i.e. WAV files). If you have an IR that is too long (and hence will take up too much CPU) then you will want to trim it in a program like Audacity.

Here are some natural IRs that you can download:

Choose shorter RT60 (like less than 1s).


Thanks !
Still interested in creating my own. @kilchhofer posted some clarinets and violin examples, i am curious how they have been recorded.
And a question : am i right to assume that the impulse duration has nothing to do with the space you’re ‘recording’, but has an impact on the overall quality ?

The process is called deconvolution. Here is a relevant question on the dsp.stackexchange which describes the basic procedure:


Impulse Response = Inverse FFT { FFT{Recorded Sound} / FFT{Original Sound} }

The Original Sound is usually a swept sine wave (20Hz to 20kHz over 30s for example). The Recorded Sound is a recording of the Original Sound played back in the space, or through the object, of interest.

The length of the IR has EVERYTHING to do with the space. Basically the length of the IR is dependent on the RT60 of the space:

which is how long it takes for reflections to die down to -60dB.


here is the thread for the mungo c0 IR files i created long ago:
the clarinet and violin files are what i call instrument body EQ’s, i didn’t create the violine, just the clarinet and that’s quite simple to do if you have an match eq software, such as izotope.

here is a link to a very very great IR collection of weird spaces, old speakers etc:

i’m very interested in equalizer IR’s. if anybody here owns sculpture, the awesome modelling synth that comes with logic then you know that there is a body eq section which does sculpt your sound more into a flute or cello or guitar sound, of course you also need the right sound to begin with.
i tried to do my own with a matched eq software, such as the equalizer from izotop ozone.
first you need a good recording of the instrument you want to match, anechoic sounds are obviously better (recorded in an anechoic chamber so no reverberation whatsoever), brian’s above posted link does contain some good recordings:
and you need a bit of white noise or a pulse (i found a spike that works very well), just something to get a very flat eq curve. i record the clarinet passage for example into the eq, record the noise into the second and match the flat curve to the clarinet, now i let a short spike or a short burst of white noise through this clarinet eq and record the outcome, this now is my IR. here’s an example i made with a female singing voice, the original audio file was her singing a scale. i put white noise through the c0 (or any convolution software) and this was the result:

suddenly it sounds like a choir since i “excite” all the frequency bumps in the IR with white noise.

here on the right side you can download sweep files and the above mentioned spike i find works best for this technique:

and here’s another example of this weird technique (the IR is a natural trumpet playing the overtone scale, input a regular saw wave):

basically this gives you very special eq shapes, i think in the rossum morpheus filter you have similar curves. now the outcome greatly depends on what you send in, just like in a resonator, some frequencies might ring, others vanish.


Great post and super weird example - thank you :smiley:

Thank you for posting these, @kilchhofer! I really enjoy your posts and your soundcloud posts as well, you get such a lovely sound of of your modular!

Oh my, that trumpet one is so beautiful. So would this IR be usable in the ER301? EDIT: Probably not, it’s over the length limit?

That’s super helpful and interesting, thanks !

super useful - thanks for the link.

i have loads of IR but 90% of them were too long

i have loads of IR but 90% of them were too long

This may be obvious, but IR files are literally just the audio of what a theoretically “perfect” impulse would sound like in the room, something like a balloon popping. If the files you have are too long, it’s because there’s a long reverb tail. You can just open the IR file in Audacity or whatever, fade out gently near the end of the window the ER-301 supports, then crop and save the file.

It won’t be exactly the same, but it might have some of the characteristics you were looking for.


no it’s very short, but not useable now for the er301 as most of the IR files i made for the c0, most of them clip the unit. i have to try stuff to find the best files and volume. on the c0 you can change the frequency of the IR, basically reducing its sample rate so i prepared them for this…bounce them with high sample rate but then reduce it in the module to lengthen and tune it, that’s why the linked files to the c0 library isn’t useable in the 301…sorry.

1 Like

i quite like this experimental convolution unit, it’s indeed ultra cpu heavy and therefore i experience quite some pops and clicks. nonetheless i love it!! here are some very sketchy and raw recordings of drone tones and ghost rhythms loaded into the unit, dry signals are filter pings and dpo strikes:


This is some really cool stuff you’re posting, @kilchhofer! You clearly have a good command of “what can I do with convolution besides reverb?”

Don’t suppose you ever do any kind of tutorials do you? :nerd:


thanks joe. well, i think of convolution as a kind of resonator which you excite with any kind of input signal, of course like any resonator it’s frequency dependant, so if you feed more high frequency content you also excite more high frequencies (if they exist in the IR), it’s a really organic way of using samples, never static and it blends in with the exciter signal. quite alot of my pad sounds in my music are made this way. feed it a rhythmic structure and the pad becomes more rhythmic, filter the input signal and the pad follows this movement etc.
tutorials are not for me, i don’t have a good camera, don’t speak english well enough and it would be terribly confusing anyway, i’m the prototypical spaghetti patcher!


I like the way you write @kilchhofer, it makes sense and is inspiring and thought provoking - the idea of IRs as resonators is fascinating and not how I had ever thought of them before!

Cheers :slight_smile:


Totally understand, @kilchhofer. :slight_smile: Even your typed notes here are helpful though!

So I have been attempting to make an IR for the first time. I’m capturing the impulse of an Alesis MIDIVerb 4, and just using some of the factory pres that I think might translate OK (e.g. no long reverb, delay, etc.)

Seems close but not quite right. In this recording (which is NOT some spectacular demo!):

  1. dry ER-301 (triangle wave)
  2. same as 1) but run through the actual MIDIVerb
  3. same 1) but this time the Exact Convo unit with the IR captured from the MIDIVerb

The impulse I created is 48k 16-bit mono. I recorded a sweep for 1s to make it, and did trim 1s impulse at the point where there appeared to be no more data (0 amplitude). So now it’s 0.397s.

Here’s the IR. (46.2 KB)

I don’t know, possibly I just picked something that can’t be captured well with convolution. All new to me. Thought I’d share in case anyone has a clue or is just interested.

EDIT: OK, so I forgot to set my latency compensation. That’s what’s responsible for the delay heard in the first track. Solved. Leaving the sample @ 1s did not solve anything and got me crackles. So through the MIDIVerb, I think I’m hearing a 1 octave down pitch shift. Through the convolution unit with this IR, here’s what Melodyne says I’m getting.

EDIT 2: So tried a different kind of preset, and the result is better. This one is a flange effect. If anything, I’d say the flange is less pronounced using the IR, but it is similar. (3.8 KB)

Looove the trumpet piece!
Thank you for sharing!