I want to look at building a DSP with Finite Impulse Response (digital) Filters to "correct" impulse response as measured.
The impulse response can be loudspeaker only, or loudspeaker plus room.
Processing power is fast and cheap now, but this has been done for at least the last 10 or so years.
If you understand FIR filters, I do not want just linear filters, but want to be able to correct phase from the measured impulse response as well.
FIR filters are now very heavily used in communications and in (classified) sonar work. If you remember, not too many years ago, a phone line could only handle a small bandwidth. That got pushed as far as it could go with with 56kpbs modems, until FIR filter DSP was applied to the circuit. The correction for the impulse response of 5 miles of twisted pair phone lines now gives us highspeed broadband connections over that same pair of wires.
Basically a very short signal is sent, and the signal that is received at the other end is carefully analyzed to see exactly how it changed (the impulse response). This is broken down into different individual effects (resonances, reflections, phase shifts) and then a DSP is programmed to precompensate for these effects. The response curve is leveled out and phase accuracy is restored by anticipating the effect of the known errors on the wire (or other transmission media) and precorrecting for them.
In this way, if a frequency band has a known phase shift through the medium, the signal gets pre-compensated with an opposite phase shift, so that this band is in phase alignment with the rest of the signal when it arrives at the other end.
I have heard it done on several loudspeaker systems, including a $60K pair of studio monitors. The effect is magical. Near perfectly flat response, and near perfect phase alignment across the entire spectrum.
The digital filter design is apparently as much art as science, so I don't think the technology is currently available to completely automate it. (Any more than a single formula or program can design the perfect analogue filters, like a crossover network).
OK, now that I have likely posted this on the wrong board (closest fit I could find - or at least the closest crowd) AND have spoken over the heads of a few AND talked down to the experts like they were 3rd graders (all of which I apologize for now) -- I need help with this. So, I am looking for recommendations. Who to speak with. What to read to get up to speed myself a bit.
Who is currently playing with this stuff?
The impulse response can be loudspeaker only, or loudspeaker plus room.
Processing power is fast and cheap now, but this has been done for at least the last 10 or so years.
If you understand FIR filters, I do not want just linear filters, but want to be able to correct phase from the measured impulse response as well.
FIR filters are now very heavily used in communications and in (classified) sonar work. If you remember, not too many years ago, a phone line could only handle a small bandwidth. That got pushed as far as it could go with with 56kpbs modems, until FIR filter DSP was applied to the circuit. The correction for the impulse response of 5 miles of twisted pair phone lines now gives us highspeed broadband connections over that same pair of wires.
Basically a very short signal is sent, and the signal that is received at the other end is carefully analyzed to see exactly how it changed (the impulse response). This is broken down into different individual effects (resonances, reflections, phase shifts) and then a DSP is programmed to precompensate for these effects. The response curve is leveled out and phase accuracy is restored by anticipating the effect of the known errors on the wire (or other transmission media) and precorrecting for them.
In this way, if a frequency band has a known phase shift through the medium, the signal gets pre-compensated with an opposite phase shift, so that this band is in phase alignment with the rest of the signal when it arrives at the other end.
I have heard it done on several loudspeaker systems, including a $60K pair of studio monitors. The effect is magical. Near perfectly flat response, and near perfect phase alignment across the entire spectrum.
The digital filter design is apparently as much art as science, so I don't think the technology is currently available to completely automate it. (Any more than a single formula or program can design the perfect analogue filters, like a crossover network).
OK, now that I have likely posted this on the wrong board (closest fit I could find - or at least the closest crowd) AND have spoken over the heads of a few AND talked down to the experts like they were 3rd graders (all of which I apologize for now) -- I need help with this. So, I am looking for recommendations. Who to speak with. What to read to get up to speed myself a bit.
Who is currently playing with this stuff?
I found a really good online (and print) book describing the fundamental process and math behind DSP filters. It goes into really good detail on how an impulse response can define exactly have a system will modify a signal.
It is called The Scientist and Engineer's Guide to Digital Signal Processing and can be read here:
http://www.dspguide.com/
At this point, I really want to find anyone playing with this technology in reference to correcting loudspeaker response, or room response.
The basics seem pretty darned simple, but I want to find out from user what hardware, specifically what DSP chips are being used.
It is called The Scientist and Engineer's Guide to Digital Signal Processing and can be read here:
http://www.dspguide.com/
At this point, I really want to find anyone playing with this technology in reference to correcting loudspeaker response, or room response.
The basics seem pretty darned simple, but I want to find out from user what hardware, specifically what DSP chips are being used.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.