There are a variety of techniques for working around the corrupting effects of environmental reflections when measuring loudspeaker radiation. I did a bit of poking around and haven’t seen the use of autocorrelation to reduce the effect of reflections. One example of using it to remove echoes in other contexts can be found here. I’d be interested if anyone knows of previous work on autocorrelation applied to speaker measurements.
Few
Few
Perhaps I should add that I've been trying to imagine whether movable "walls" (light weight but acoustically reflective barriers) might be used to block room reflections from reaching the mic. The movable walls would create their own reflections, of course, but if they are sufficiently clean--meaning not corrupted by a thousand reflections from whatever junk happens to be cluttering the room--the reflections might be significantly reduced via autocorrelation. Plus moving the walls would change the arrival time of their reflections at the mic, and help distinguish the reflections from the direct sound. If the approach worked perfectly, the results would be the same independent of wall-mic distance, so there would be a way of determining over which frequency range the measurement results are to be believed. Barriers of practical size won't work down to 20 Hz, but maybe they could at least be another way to gain a few msec of clean impulse response.
I've tried very thick walls of absorbing material around the mic but never got very far with that approach. I've been left wondering whether an "if you can't beat them, join them" approach might be better. Admit there will be reflections, but make them sufficiently clean and identifiable so that they can be more readily removed.
I haven't tried this in practice yet, but I was hoping someone might be able to help me avoid repeating work others have already done.
Few
I've tried very thick walls of absorbing material around the mic but never got very far with that approach. I've been left wondering whether an "if you can't beat them, join them" approach might be better. Admit there will be reflections, but make them sufficiently clean and identifiable so that they can be more readily removed.
I haven't tried this in practice yet, but I was hoping someone might be able to help me avoid repeating work others have already done.
Few
Hi,
https://www.bbc.co.uk/rd/publications/rdreport_1995_04
In 'basic assumption' section you should find interesting ROT about size of rederecting panel wrt freq of interest.
Autocorrelation could be of interest but i see issue in domestic rooms: nature of wall, need for very accurate location of mic and source,... maybe with AI implied it could lead to something usable without too much issues?
After all this is kind of similar to solution D. GUNNESS use on Fulcrum's to correct the issues of protubering horn in coax in my view.
Maybe trying to correct impluse response could be easier to implement? ( preconditioning to counteract the impulse distortion).
https://www.bbc.co.uk/rd/publications/rdreport_1995_04
In 'basic assumption' section you should find interesting ROT about size of rederecting panel wrt freq of interest.
Autocorrelation could be of interest but i see issue in domestic rooms: nature of wall, need for very accurate location of mic and source,... maybe with AI implied it could lead to something usable without too much issues?
After all this is kind of similar to solution D. GUNNESS use on Fulcrum's to correct the issues of protubering horn in coax in my view.
Maybe trying to correct impluse response could be easier to implement? ( preconditioning to counteract the impulse distortion).
Last edited: