Jonathanese
Registered
Thread Starter
- Joined
- Oct 23, 2020
- Messages
- 7
More
- Front Speakers
- Adam T7V
- Other Speakers or Equipment
- Custom 40W DML speakers
- Streaming Equipment
- Rode NT1a
- Streaming Subscriptions
- Scarlett 2i2 Gen3
This isn't so much a "how to do" but rather an an idea for a potential feature.
I've started to use REW's filter FIR output in EqualizerAPO and FL Studio using Convolver (I've also used Pro-Q 2 which has great response-matching as well.) Basically, I use the FIR as a baseline and add additional tweaks manually. Mostly I figure doing a single convolution is likely easier than running and managing 20 discrete filters.
So I was thinking, since I'm not working with discrete filters in these cases, what would it take to generate a corrective impulse response directly from the measured impulse response? I would imagine you could get this from measuring an impulse response, then taking an ideal impulse, and deconvolving it with the measured response. That way you get a response that, when convolved with the room behavior, gets you close to an ideal impulse response. Albeit you would still want a bit of blurring as you'll probably just make things worse everywhere else. But at least in this way, it is able to pull out peaks it might not be able to cover using 20 filters.
I'd say probably the big issue would be how accurate a digital deconvolution could be. I took feedback control systems but not DSP classes so I'm not sure if that breaks down.
Might be something I'll experiment with on my own. But if it works out, it could be a cool feature that might just look like a "generate inverse impulse response as WAV" in the options menu.
I've started to use REW's filter FIR output in EqualizerAPO and FL Studio using Convolver (I've also used Pro-Q 2 which has great response-matching as well.) Basically, I use the FIR as a baseline and add additional tweaks manually. Mostly I figure doing a single convolution is likely easier than running and managing 20 discrete filters.
So I was thinking, since I'm not working with discrete filters in these cases, what would it take to generate a corrective impulse response directly from the measured impulse response? I would imagine you could get this from measuring an impulse response, then taking an ideal impulse, and deconvolving it with the measured response. That way you get a response that, when convolved with the room behavior, gets you close to an ideal impulse response. Albeit you would still want a bit of blurring as you'll probably just make things worse everywhere else. But at least in this way, it is able to pull out peaks it might not be able to cover using 20 filters.
I'd say probably the big issue would be how accurate a digital deconvolution could be. I took feedback control systems but not DSP classes so I'm not sure if that breaks down.
Might be something I'll experiment with on my own. But if it works out, it could be a cool feature that might just look like a "generate inverse impulse response as WAV" in the options menu.
