I am in the middle of configuring BruteFIR for bass redirection and digital room correction duties on my 5.1 system, but have run into a bit of an obstacle.
I generated filters using the fir1 function of Octave, having 8192 taps and crossing over at 60Hz. I then loaded these into brutefir: the highpasses feeding my main speakers and the lowpasses feeding through to the sub.
Next I measured the distance of each speaker from the listening position, and worked out the delays necessary (in samples) to time align the speakers. I then inserted these delays into the output section of the brutefir config.
With the system now setup I did log freq sweeps on the left and right channels (xovered through brutefir to the sub) and ran DRC on the resulting pcm impulse responses. Alarm bells sounded however when I noticed that drc found the impulse response centre of the right speaker's impulse response almost 7000 samples different to that of the left speaker's impulse response! usually I find that they are identical since I make sure the microphone is exactly equidistant from the left and right speakers.
I then checked out the frequency response of the two recorded impulse responses using cooledit and found something very surprising: please see attached file.
This has caused much head scratching, if anyone has any experience of this kind of thing or has any suggestions I'd be very grateful to hear from you.
Many thanks,
Chris Birkinshaw
ps. brutefir config is at http://postbox.org.uk/drc/brutefir_config060105 (DRC coeffs are listed but disabled for now with a "-1" in the relevant filter section)
I generated filters using the fir1 function of Octave, having 8192 taps and crossing over at 60Hz. I then loaded these into brutefir: the highpasses feeding my main speakers and the lowpasses feeding through to the sub.
Next I measured the distance of each speaker from the listening position, and worked out the delays necessary (in samples) to time align the speakers. I then inserted these delays into the output section of the brutefir config.
With the system now setup I did log freq sweeps on the left and right channels (xovered through brutefir to the sub) and ran DRC on the resulting pcm impulse responses. Alarm bells sounded however when I noticed that drc found the impulse response centre of the right speaker's impulse response almost 7000 samples different to that of the left speaker's impulse response! usually I find that they are identical since I make sure the microphone is exactly equidistant from the left and right speakers.
I then checked out the frequency response of the two recorded impulse responses using cooledit and found something very surprising: please see attached file.
This has caused much head scratching, if anyone has any experience of this kind of thing or has any suggestions I'd be very grateful to hear from you.
Many thanks,
Chris Birkinshaw
ps. brutefir config is at http://postbox.org.uk/drc/brutefir_config060105 (DRC coeffs are listed but disabled for now with a "-1" in the relevant filter section)
