Re: Seeking Help Developing a Crude Room Correction Algorithm



gener wrote:

Hi all, first time poster here, and was hoping to get some advice on class
project. What I'm trying to do is to correct the impulse response of my
room by filtering the stimulus signal so that the new IR would be "flat".
Then, hopefully sending out the new stimulus signal would result in a flat
frequency IR тАФ this is crude (i'm a beginner), and I'll probably avoid
dealing with phase at this point.

It is a trivial problem to equalize the response of a room for any particular location of a speaker and a mike in this room. However it is not a trivial problem to equalize a room in general.

If I want to scale the frequency bins of the stinmulus signal by the
inverse of the magnitudes of the IR, any suggestions on how to make sure
the bins in the stimulus and the IR are the same as I take the FFT of
each?
> Basically I want to convolve (.*) the ffts of the stim and the IR and then
> take the ifft of that result.


With FFT, it is very likely that you will run into zero divided by zero uncertainty. Use an adaptive filter such as LMS algorithm instead.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com


.

Relevant Pages

  • Re: Frequency shifting for software defined radio
    ... I've mostly been doing the magic in the time domain and using FFT ... I've tried FFT fast convolution filtering and got it working. ... Just moving the bins in the FFT output... ... Tauno Voipio (OH2UG) ...
    (comp.dsp)
  • Re: Frequency shifting for software defined radio
    ... I've mostly been doing the magic in the time domain and using FFT ... I've tried FFT fast convolution filtering and got it working. ... Just moving the bins in the FFT output... ...
    (comp.dsp)
  • Re: Step response from adding impulse responses together
    ... non-zero impulse response, ... it from the original filtering formula. ... I would assume that there's plenty of literature on various ways to do ...
    (comp.dsp)
  • Re: Linear Interploation Algorithms
    ... you do a lowpass _before_ the rescaling, to avoid aliasing. ... But you will perhaps want the lowpass to avoid the aliasing effects. ... A perfekt low pass equivalent to your fft approach would be sin/x ... Scaling and filtering can folded and performed in one step. ...
    (comp.arch.fpga)
  • Re: Step response from adding impulse responses together
    ... non-zero impulse response, ... it from the original filtering formula. ... I would assume that there's plenty of literature on various ways to do ...
    (comp.dsp)