[Csnd] Creating IRs: where is the problem?

The Csound Mailinglist
1

Hey hey,
I have written some code and a full process, as I understand it. It's here:
https://www.dropbox.com/scl/fi/u8ek20e8cld6pioqajb1g/make_ir.zip?rlkey=smr1pkpd1dkwao79cf7o9of1i

Short guide: you can just run the .csd files in alphabetical order, they are numbered. The first one create the sweep sound, the inverse sxweep and the necessary scaling factors. I think, I don't even need to store that table.

The second orchestra records a sine sweep through reverbsc.

The third file might be the really bad one. It convolves the recorded sweep with the inverse sine sweep.

The fourth file tries the result with fox.wav and it sounds... Creative, but nothing like a reverbsc would sound.

I'm sure, I misunderstood part of the process. Do I perhaps need to reverse the recorded sweep or use a different convolution method with that inverse sweep?

Best wishes,

Jeanette

2

Hey hey,
I have seen the errors of my ways and taken down the example file. I can fix it now.

Just one question, for a bertter workflow, can someone post an instrument/code snippet to perform convolution in an i-time loop. Assume I have the tables iInvSweep and iRecording. The recording table holds the recording of the sine sweep played in the room. If convolution could be performed using the basic fft functions, the whole process wouldn't rely on actual running time and might all around be a little simpler to use from the outside.

Thanks and best wishes,

Jeanette

Jeanette C., May 6 2026: