Home Theater Forum and Systems banner

1 - 7 of 7 Posts

·
Registered
Joined
·
9 Posts
Discussion Starter #1
Hello all...

I am trying to do something that REW is not specifically designed to do...but I can't imagine why I can't...

I am using the very nice OpenDRC unit from MiniDSP and I need to generate a FIR filter to load in the OpenDRC. I am using the procedure described on MiniDSP's website that incorporates the use of REW to create a FIR filter through exporting impulse.wav files and using the Python-based porc.py program by Mason Green, to create an inverse filter. I use Sox to translate the .wav file to a FIR-based .bin file that the OpenDRC accepts into memory. The .bin file loads nicely into the OpenDRC and corrects my crazy speakers for a nice flat response in my room. All is good...now I need to take it to the next level.

I want to use the OpenDRC to filter the stereo microphone feed from a home-made 3-D manikin recording dummy. Therefore, I need a filter with a Head Transfer Reference Function (HTRF) function of the manikin dummy. A calibrated "flat" response target is NOT satisfactory for a HTRF. I need to measure all the bumps and blemishes of a reference sound field using a broadband mic and compare it to the response of the microphones used in the ears. A proper HTRF needs a FIR filter that represents the DIFFERENCE between a reference response of a broad-band microphone outside the head, and the same microphone located in the ear (head) of the manikin dummy.

Can anyone tell me if I can create a "calibration" curve (reference microphone) of a wildly non-smooth response in REW and use it to "measure" the HTRF of a microphone placed in the ear?

I have tried this twice with less than desirable results. Maybe I am doing something wrong. I am hoping that I can specify a .cal file with the response of the reference microphone...maybe not. Can someone tell me if there is a trick to getting a calibration curve that I want so I can make relative comparative HTRF measurements?

Thanks

Phil Mc Can D
Distorted name for anti-archival/search purposes
 

·
REW Author
Joined
·
6,711 Posts
You can load pretty much any response you like as the soundcard and/or mic/meter cal file and REW will subtract it from your measurement, just export the reference measurement you want to use as a text file from REW then load it as the cal file. There is a 20dB boost limit when cal data is applied, which you can override if necessary by unticking the box on the Analysis preferences tab under Frequency Response Calculation.
 

·
Registered
Joined
·
9 Posts
Discussion Starter #3
John,

Thanks for the prompt reply. I will try your suggestions re "de-limiting" the reference cal soon.

Another question is that for a proper HTRF, I would really prefer a "Diffuse" sound field presentation. The easiest way to measure in a diffuse field is to place the dummy in a highly reverberant cement-walled room and allow a steady-state signal to bounce around for a while.

How can I implement a multi-directional diffuse field response if the only way I see to generate a filter in REW is using a mathematically "derived" impulse response (which I assume would be largely a direct wave ignoring reflections)?

Phil Mc C.And
 

·
REW Author
Joined
·
6,711 Posts
Can't help there, sorry - been quite a few years since I played with HRTF and EQ. If I recall correctly an old Stax ED-1 I measured many years ago had an equalisation curve that was pretty much the inverse of the dummy head measurement, so was basically equalising to a flat response.
 

·
Registered
Joined
·
9 Posts
Discussion Starter #5
John

ok, so you can't help with the diffuse-field stimuli, so how about the next best thing....Is there a way for REW to take several measurements and combine them together to derive an "averaged" impulse? This is to allow me to rotate the dummy head several degrees at a time, run independent measurements and combine them in a pseudo-diffuse field that I can generate a FIR filter from?

Just askin'


Phil M cCan.d
 

·
REW Author
Joined
·
6,711 Posts
You can average measurements on the All SPL graph (open up the controls for that graph to see the trace arithmetic options). If you need an averaged impulse response, you need to make sure the measurements are time aligned. You can then use the (A+B)/2 option in trace arithmetic to product pairwise averages, then average the pairs etc to work down to a single averaged result (trace arithmetic only works on pairs of traces, hence the somewhat clumsy approach needed).
 
1 - 7 of 7 Posts
Top