I'm working on a new audio project, and to start off, I need a filter that allows me to manipulate both frequency and phase response. Can I do both things with one filter? What kind of filter would that be?

It's called parametric eq. Look it in sound editors(cool edit has it) and be sure to check with an impulse that it has no PRE ringing, only post(that it is a minimum-phase filter)

erm, parametric eqs don't let you control the phase response.

You need a convolver or Finite Impulse Response (FIR) filter.

If you have no idea where to start, you could start by finding a filter design package. No, I don't have any suggestions, because I use the one built-in to MATLAB.


If you want to define an arbitrary amplitude and phase response, there is a window FFT technique (for designing an FIR filter - you can use FFT overlap+add or straight convolution to do the actual filtering) which you can find in all digital audio processing textbooks. That's the place to go.


Depending on what you want, it can be much more efficient to split the filter into two simpler part, which could be (but don't have to be) amplitude and phase. A filter which only changes the phase of various frequencies is called an All-pass filter.


I hope this gives you enough information to start looking. Really, you need a good textbook. Visit a library or a book shop!

Cheers,
David.

Gday..



yes.. and recomended package is waves: waves 4 all..
do what you need.. + and what you ever would need.. (DX/VST/RTAS)

But, can you design a filter with arbitrary phase response with Waves? I doubt it, but I don't know, since I haven't tried it yet.

Ciao..

hmmz..
i was little to fast on this.. although i seem to recall waves 4. do have a filter
who does that.. check it out.. to be sure.. http://www.waves.com

anyway.. check out mathworks: http://www.mathworks.com/products/signal/f...functions.shtml

and here is a paper i found on designing IIR filters with arbitrary freq. response.
http://cmc.rice.edu/docs/docs/Var2001May5O...5ONTHEDESIG.pdf

Thanks for the replies. It turns out that the transformations I need to apply are already encoded in the form of impulse responses, so I just need to use those for my filter kernel...