Room Correction with PEQ

[Baseballbat]

pos,

of course you'll have pre-ringing with every linear phase filter. But as long as the filters are short, it doesn't matter, and you can still have steep (enough) slopes.

 

[pos]

Ok, but then that should read "But as long as the slopes are not (too) steep, it doesn't matter, and you can implement them with short FIRs."

 

[jmbee]

Hi jmbee, sorry for the delay to respond to your post. Looks great! Would you be able to share a bit more on how you achieved your results? Cheers, Mitch

Hello Mitch,

This result comes from using a pc based multi way crossover like discribed by Thierry here:

http://www.diyaudio.com/forums/pc-based/223805-easy-fir-crossover-pc-based-drc.html

Each H-P is measured then the response is shaped (equalized / filtered ) using rePhase to synthetise the convolution impulses, (2 x 3 ways ) and VSThost + VSTconvolver makes the all run.

Also is inserted an independant global equalizer to tame low frequency peak from the room modes.

crd

An externally hosted image should be here but it was not working when we last tested it.

 

[bohdan1232000]

DRC package contains Kirkeby inverse routine. All other FIR correction software contains similar routine.

Hi Barleywater,

Sorry for this belated response, I have just been very busy lately.

Your statement quoted above is not quite correct. I had a quick look at the Kirkeby method and made some comments in

http://www.bodziosoftware.com.au/Comments_On_Kirkeby_Method.pdf

Best Regards,
Bohdan

 

[markus76]

^
Thanks Bohdan!

 

[Barleywater]

Bohdan,

I read your analysis many years ago.

Your approach to analysis is limited. Bandwidth limited.

Much better to generate inverse across entire driver bandwidth, then apply desired high pass and low pass filters to constrain inverse function to range of interest.

 

[gedlee]

Very interesting papers. Thanks

 

[Barleywater]

Another very good paper and it's references:

http://audioroundtable.com/misc/Mueller.pdf

Regarding Kirkeby, pay attention to sec 1.7, and 4, especially comparison of 4.1 and 4.2.

Kirkeby with suitably small regularization parameters performs 4.2