Create your own impulse responses

I’m definitely no pro response maker, but have made some in the past using ghetto skills & equipment and found it great fun…

Procedure:

  1. You need a computer program to produce a sweeping sound from low to high frequencies.

For this I use a script in Praat on windows, get the freeware program here: Downloading Praat for Windows

Go to top left of Praat Objects window gui; ‘Praat\New Praat script’ and insert:

startFreq=10  
endFreq=12800  
lnRange=ln(endFreq/startFreq)  
periodT=10  
multFact=lnRange/periodT  
Create Sound... ExpSweep 0 periodT 44100 exp(multFact*0.5*(x-periodT))*sin(2*pi*startFreq/multFact*exp(multFact*x))   

Press run from the script window. This will create a sweep of 10 seconds going from 10 to 12800 Hz. You need to watch out with setting the end frequency, setting this to high can blow up speakers so I’ve heard. Horror stories or fiction, you’ve been warned and I can’t be held accountable :slight_smile:

The generated sweep can be found in the Objects list, for playback, analyzing or saving. Go to top of Praat Objects window and select Save\Save as Wav file.

You now want to open the generated sweep in a wave editor and apply some minor fade-in & outs to prevent a burst of higher energy at the onset and end of the sweep. Additionally you can add silence at the end of the sweep, enough silence for capturing the expected reverb-tail that is to be recorded. This is relevant when looping the generated sweep, moving around with the microphone in space, trying out different positions.

  1. Ideally you want to output the sweep-sound through 1 speaker with a flat as possible frequency response. Meaning an even signal across all frequencies. But as this is utopia tech speak, use whatever speaker you have, can be fun for experimentation. Also keep in mind the time of day when recording, for example if you want to capture a space near a busy city street, traffic noise bleeding into the recordings etcetera, you might want to record late at night when traffic has died down. Also be aware of the position of the speaker / microphone inside a space, have the speaker near a wall and bass frequencies will get emphasized (proximity effect).

  2. You’ll need a recording device, ideally an omni directional microphone to capture the sweep + ambiance of the room/space. But I’ve made responses using a directional cheap ass Zoom H2 recorder. These handy recorders have different pick up patterns you can experiment with, also auto-mix to mono option. Make sure the recording levels don’t clip when recording.

  3. When having recorded some spaces, you’ll need to convolve the recorded sounds with the reversed version of the original sweep to create the actual impulse responses. This convolving business can be done in Praat.

Top of Praat Objects window, select ‘Open\Read from file’ and select a reversed version of the used sweep and the recordings. Using ctrl or mouse drag selecting, select the reversed sweep and a recorded result and press ‘Combine -’ in the Synthesize options of the Praat Objects gui. Now select Convolve…

I can’t remember what settings should or shouldn’t be used in the convolving window, I think using the defaults, or normalize in the Amplitude scaling works just fine before pressing Apply. A new object will be created in the Praat Objects list, save it as .wav file and open it in Renoises convolver unit :drummer:/> .

Remember:

Here’s some I once did for school of a place I lived in (generated praat impulse response and reversed version also included / note some of these are very bassy, adjust gain & wet mix!);

https://www.mediafire.com/?fo06zecjn8lludz

Sounds cool, but i wonder if it has to be made with this particular sweep of yours or does it work with other sounds aswell? Could i make the sweep in renoise? Or am i far from understanding how this works? :o

I may want to try that on my acoustic guitar body one day… Wondering if i could get some more warmth out of my audio using an IR.

Praat is deeeeep , whew …can get some really interesting stuff out of that program , the vocal traction stuff alone is mindblowing
But a simple sine sweep can be done with any program for that matter ,vst or even an audio editor .