Yes heard a pair briefly. Sounded good too me but it was not in the best acoustic ever ( a broker's warehouse...). I think it is loudspeakers which are (were?) at ICP studios in Belgium but i'm not sure, salesmen are salesmen and you cannot always believe them...
In comparison to the same kind of monitors i was exposed to at that time (big Kinoshita/Atc/Genelec) it was into same league. I remember have been impressed by the clean high SPL they produced and they were less arsh in the high end than the RM-4 i knew well at that time. This day we heard ATC, Pmc and Boxer systems too.
I could have been pleased with all of them... it ended with a smaller ATC ( scm110a) in front of me for 2 years.
Not heard the Bluehorn but spent hours with HD-1. I liked them a lot, better than the 1031 which were 'standard' at that time (with NS10). They had a kind of 'clarity' to them.
Spent all day reading this thread
Wow what an awesome read , I said I wouldn’t interrupt back on the first couple pages…. Now 28pages later
Man that virus guy has never used an OpenDrc before that’s so obvious… internet expert there🙃
so , what a bunch of good info…..
So I might have said it before earlier, so sorry if I’m repeating…
So one thing I’ve noticed with a single correction vs multiple corrections is they sound about the same once either completed and all things being equal as far as mag and phase response goes ,
But here’s one tidbit…. I have done driver crossovers near the speaker and achieved flat mag and phase across a multi way…
I sum them together, and at the LP there’s phase wraps! Yes!! Having REW remove tof delay and center ir using same as drivers measurements….
So the room truly does add its own wraps and speakers add there own phase issues, so I had to do a 2nd correction over all of the drivers in a gloabal L and R phase correction.
So , if someone is doing a single fir for everything, it would have to be at LP…. If a fir was available for each driver would it be better to do it at LP or the driver,
Wow what an awesome read , I said I wouldn’t interrupt back on the first couple pages…. Now 28pages later
Man that virus guy has never used an OpenDrc before that’s so obvious… internet expert there🙃
so , what a bunch of good info…..
So I might have said it before earlier, so sorry if I’m repeating…
So one thing I’ve noticed with a single correction vs multiple corrections is they sound about the same once either completed and all things being equal as far as mag and phase response goes ,
But here’s one tidbit…. I have done driver crossovers near the speaker and achieved flat mag and phase across a multi way…
I sum them together, and at the LP there’s phase wraps! Yes!! Having REW remove tof delay and center ir using same as drivers measurements….
So the room truly does add its own wraps and speakers add there own phase issues, so I had to do a 2nd correction over all of the drivers in a gloabal L and R phase correction.
So , if someone is doing a single fir for everything, it would have to be at LP…. If a fir was available for each driver would it be better to do it at LP or the driver,
Oabeieo said:
To me the science (and mathematics) behind everything are more important than subjective opinions that say something like "it sounds good, so I like it". And, of course, it is always easier to purchase something which is what consumers do. But I'm not a consumer, I'm an engineer.
Yes, very low frequencies, combined with IIR highly increase the requirements on calculation accuracy.
Regards
Charles
Let me make sure I follow you....
when you make crossovers near the speaker, are they complementary linear phase? Where you use a channel of openDRC for each driver?
If so, for room-type speakers, where measurements are hopefully made at somewhere approaching listening distance, there should be no major phase wrap change at LP, once ToF is removed.
Phase variations can occur, but they should be confined to what happens from tuning at one distance vs another. Iow, the geometric focus distance.
Measurements are advised to be taken in the far-field whenever possible.... the acoustic definition of far-field where phase between drivers has settled down, and all drivers'SPL falls with distance at inverse square law.
But If you're making driver measurements right at the speakers in the car, I can see how everything changes at LP...majorly.
I KNOW NOTHING about car audio!
Yeah so it goes
3 2x4hds for 12ch
Mmm close range to speaker and make response flat. Use 10 peq iirs per ch
Each output has fir…. Then a full complimentary crossover system wide. Measurements done at driver, 3 close range measurements (averages and summed)sent to rephase for phase flattening and crossover activation using 1024 taps per driver and 2048 taps for the sub on a 4 way.
After each speaker has been corrected , they all get summed together at LP. Basic MMM averages to get a magnitude build for a flat response (except sub gets layered at 10db above). At this point things don’t sum flat… it’s bumpy……the crossover are then forced to match at the planned -6db area using eq done in the OpenDrc (6iirs)
After that fir iterations of eq to make response flat with minimum phase eq and using mmm averages 16k pink PN. Make response flat in rephase and send back to validate…. Takes about two tries before magnitude is flat..
Then take 9 sweeps around my head and time align and average and send to rephase for final phase correction using measured phase. Done … and sounds fantastic..
I ignore the response issues from the sweeps and just do phase only, the mmm rta made final response flat and much better tone. I also used measured phase vs excess phase. It sounds a little better that way… excess phase and measured phase are almost the same thing anyway… the excess phase was like 100hz difference from measured phase and had the exact same shape and wraps.
So the OpenDrc phase ended up having 3 wraps….. the room had to have added them… idk how, there wasn’t enough eq to make wraps except near 12k where I don’t bother anyway with at LP….
The near field groups everything was corrected. All drivers made pancake flat throughout the entirety of its range, and some compensate filters applied where taps allowed…. The sub near field has no tilt and was flat in phase, at LP it tilted down from DC to 100hz (I thought that was normal behavior and left it)
A car isn’t different from a very small room and your path lengths are unequal….. it really isn’t. Things behave pretty much the same as a home, just far more issues to deal with. In fir that stuff is wiped away pretty good and easily. The only thing that can’t be fixed is deep dips that are room modes…. They have to be ignored and treated as a flat response. Just pretend it’s flat response. And do phase correction as if it had no dip. I have found it sounds better like that and I think the energy is still there in the room. So somewhere in the room it’s heard.
All speaker corrections are done at driver and OpenDrc is done separately left and right at LP.
This is the only way to get all the way through the correction without seeing de-correlation and loosing the center.
If I tried doing LP crossovers and driver eq the eq portion de-correlated and I loose center imaging. So if anyone can speaker to de-correlation then I’m listening.
My gut tells me most home audio ppl are only doing one driver and applying there fir to both drivers. That will keep them coherent. I haven’t seen anyone post pics or screenshots of 2 rephase screens…. Only 1 ….
3 2x4hds for 12ch
Mmm close range to speaker and make response flat. Use 10 peq iirs per ch
Each output has fir…. Then a full complimentary crossover system wide. Measurements done at driver, 3 close range measurements (averages and summed)sent to rephase for phase flattening and crossover activation using 1024 taps per driver and 2048 taps for the sub on a 4 way.
After each speaker has been corrected , they all get summed together at LP. Basic MMM averages to get a magnitude build for a flat response (except sub gets layered at 10db above). At this point things don’t sum flat… it’s bumpy……the crossover are then forced to match at the planned -6db area using eq done in the OpenDrc (6iirs)
After that fir iterations of eq to make response flat with minimum phase eq and using mmm averages 16k pink PN. Make response flat in rephase and send back to validate…. Takes about two tries before magnitude is flat..
Then take 9 sweeps around my head and time align and average and send to rephase for final phase correction using measured phase. Done … and sounds fantastic..
I ignore the response issues from the sweeps and just do phase only, the mmm rta made final response flat and much better tone. I also used measured phase vs excess phase. It sounds a little better that way… excess phase and measured phase are almost the same thing anyway… the excess phase was like 100hz difference from measured phase and had the exact same shape and wraps.
So the OpenDrc phase ended up having 3 wraps….. the room had to have added them… idk how, there wasn’t enough eq to make wraps except near 12k where I don’t bother anyway with at LP….
The near field groups everything was corrected. All drivers made pancake flat throughout the entirety of its range, and some compensate filters applied where taps allowed…. The sub near field has no tilt and was flat in phase, at LP it tilted down from DC to 100hz (I thought that was normal behavior and left it)
A car isn’t different from a very small room and your path lengths are unequal….. it really isn’t. Things behave pretty much the same as a home, just far more issues to deal with. In fir that stuff is wiped away pretty good and easily. The only thing that can’t be fixed is deep dips that are room modes…. They have to be ignored and treated as a flat response. Just pretend it’s flat response. And do phase correction as if it had no dip. I have found it sounds better like that and I think the energy is still there in the room. So somewhere in the room it’s heard.
All speaker corrections are done at driver and OpenDrc is done separately left and right at LP.
This is the only way to get all the way through the correction without seeing de-correlation and loosing the center.
If I tried doing LP crossovers and driver eq the eq portion de-correlated and I loose center imaging. So if anyone can speaker to de-correlation then I’m listening.
My gut tells me most home audio ppl are only doing one driver and applying there fir to both drivers. That will keep them coherent. I haven’t seen anyone post pics or screenshots of 2 rephase screens…. Only 1 ….
Last edited:
Man I came in with rookie experience with fir and it was boldly apparent you have no experience otherwise you wouldn’t have said what you said…. I can’t assume your dumb, that obviously isn’t the case, I think your arguments with these guys from a reader standpoint was way off base and just lacked experience with fir….
I believe your an engineer, so what? That doesn’t give you fir experience in acoustics
Oabeieo said:
Could you please point out which arguments, exactly you were referring to ?
Oabeieo said:
That means that I may not see / use technology the way other people do, for example the usage of OpenDRC.
All of them, mostly
You tried relating non linear attributes from a speaker and mixing the term “linear” with linear phase systems….
Linear simply means flat in a simple term. There two separate distinct functions. There is no point in bringing it up again on this thread and wasting an otherwise fantastic read packed with knowledge and sheer experience in audio. These guys are the very top of the audio food chain. Respectfully please don’t interfere with a good thread. I’m super suprised you got the attention you did. That shows the kindness people have as humans. Just be wrong on this now please and go buy some minidsp and start doing. You’ll be pleasantly astonished how much magic fir brings to audio.
In car audio competition a big deal is “bass to the front” and getting the rear sub to sound like it coming from the front. I imagine in a home the sub can be anywhere in the room and sound like it coming from the mains…. When ALL the frequencies are in time with all the speakers, the illusion is just magical…. It’s not some trickery like using delays to hass a system to get an effect as a very narrow band, this does the entire spectrum. The speakers power compression is a non issue… that has absolutely nothing to do with this. Nobody in here is turning their system up so incredibly loud that power compression is even an issue… even if it was it’s a completely separate thing and has nothing to do with the filter…. Who in their right mind is building filters to compensate for power compression.. that’s just flat out abuse and dumb. The speakers I have in my system cost $12,000 for just the drivers. I guarantee the person that designed the focal utopias would absolutely cringe at the thought of somebody turning up the speaker and driving it into power compression…
So if you would please let this thread get back on course with the topic of the thread. I would greatly appreciate it because I spent my entire day off yesterday reading this thread, and most of it was very exciting and interesting debate which I love to read. The nonsense however please just stop.
You tried relating non linear attributes from a speaker and mixing the term “linear” with linear phase systems….
Linear simply means flat in a simple term. There two separate distinct functions. There is no point in bringing it up again on this thread and wasting an otherwise fantastic read packed with knowledge and sheer experience in audio. These guys are the very top of the audio food chain. Respectfully please don’t interfere with a good thread. I’m super suprised you got the attention you did. That shows the kindness people have as humans. Just be wrong on this now please and go buy some minidsp and start doing. You’ll be pleasantly astonished how much magic fir brings to audio.
In car audio competition a big deal is “bass to the front” and getting the rear sub to sound like it coming from the front. I imagine in a home the sub can be anywhere in the room and sound like it coming from the mains…. When ALL the frequencies are in time with all the speakers, the illusion is just magical…. It’s not some trickery like using delays to hass a system to get an effect as a very narrow band, this does the entire spectrum. The speakers power compression is a non issue… that has absolutely nothing to do with this. Nobody in here is turning their system up so incredibly loud that power compression is even an issue… even if it was it’s a completely separate thing and has nothing to do with the filter…. Who in their right mind is building filters to compensate for power compression.. that’s just flat out abuse and dumb. The speakers I have in my system cost $12,000 for just the drivers. I guarantee the person that designed the focal utopias would absolutely cringe at the thought of somebody turning up the speaker and driving it into power compression…
So if you would please let this thread get back on course with the topic of the thread. I would greatly appreciate it because I spent my entire day off yesterday reading this thread, and most of it was very exciting and interesting debate which I love to read. The nonsense however please just stop.
A lot going on 🙂
So, it take it you have OpenDRC's being used on top of the three 2x4HD's ? The 2 ch, 6144 taps per ch opendrc-di's ?
Yeah, IIR EQ's shouldn't make any full wraps.
Gotcha. I think that's the best way to do global correction...ends up being all room/car.
Yeah, I guess most people are trying global, and not FIR driver by driver. More work, more parts......
Here's a quick 'tune to a spot', driver-by-driver, done with FirD on a 4-way main,........ each driver with it's own FIR file of course.
Looks so clean, it looks electrical, not acoustic huh? Tune to a spot ought to be against the law Lol
That is an acoustical measurement my goodness…that is what I want!!! Wow that’s impressive
And that wavelet wesayso posted in post 29
That’s is amazing,…
And yes , OpenDrc is upstream of the 2x4hds and 6144 taps at 48k , that’s where the real power is…
The 2x4hds by themselfs make a great sound, but finishing the phase and toning the system in the OpenDrc really pulls it all together….
I used to have the Dirac box where the OpenDrc is and it was good, the OpenDrc is much better. I subsequently moved the Dirac box to my shop audio system. And am soon selling it to do a flex with manual fir.
If I was up against a choice of separate firs per driver or a summed fir for everything, it’s a hard choice
If someone could truly build a minimum phase crossover that performance is exact and sums perfect I think a gloabal fir would be better. Altho that’s extremely difficult to do… and because of that I would have to vote separate firs.
As soon as I get paid I’m going to try this firD….. I want that response like yours… I love RePhase, a little more automation would be nice. Just a little to help me get that. And a way to do iterations…
In rephase to do iterations, you have to bypass everything when you import a 2nd validation measurement, make the changes , then activate all banks and linearizarions. It would be cool to not do that because sometimes I forget where I’m at….
And that wavelet wesayso posted in post 29
That’s is amazing,…
And yes , OpenDrc is upstream of the 2x4hds and 6144 taps at 48k , that’s where the real power is…
The 2x4hds by themselfs make a great sound, but finishing the phase and toning the system in the OpenDrc really pulls it all together….
I used to have the Dirac box where the OpenDrc is and it was good, the OpenDrc is much better. I subsequently moved the Dirac box to my shop audio system. And am soon selling it to do a flex with manual fir.
If I was up against a choice of separate firs per driver or a summed fir for everything, it’s a hard choice
If someone could truly build a minimum phase crossover that performance is exact and sums perfect I think a gloabal fir would be better. Altho that’s extremely difficult to do… and because of that I would have to vote separate firs.
As soon as I get paid I’m going to try this firD….. I want that response like yours… I love RePhase, a little more automation would be nice. Just a little to help me get that. And a way to do iterations…
In rephase to do iterations, you have to bypass everything when you import a 2nd validation measurement, make the changes , then activate all banks and linearizarions. It would be cool to not do that because sometimes I forget where I’m at….
In that case, you're just being judgemental, as all my statements have enough science behind them. Now, if that science doesn't matter to some, then that's not my fault.
No, not at all, the "linearity" I mentioned is necessary for any equalisation mentioned in this thread (FIR / IIR) to work perfectly. Now, anyone who doesn't consider non-linearity while applying EQ is just assuming a linear system.
It takes some mathematics to understand the connection between linearity and equalisation. The math of non-linear systems is more advanced than that, but can be learnt if desired.
Like I said earlier, that kind of response where you rush to buy MiniDSP / OpenDRC is that of a consumer, which I'm not. So, I think I already answered this above, even before you put in this comment !!
There's absolutely no magic within any DSP and what you believe to be magic is exactly what I termed as a "subjective opinion" in my earlier message. You just didn't seem to get it.
You expect everyone to have the same opinion (and equipment) as you and when that doesn't happen, you don't seem to like it !!!!
Now, if you've already decided that I'm talking nonsense, then is it even possible for me to convince you otherwise ?
All the best.
We take measurements, and apply filtering , and it sounds good, then it measured good.
All the other non linearities are different subjects ….. I think that’s all I was trying to say…
I’m not being judgmental, I’m very much learning this also.
I understand your points, and that’s a different subject all together. That’s all I was trying to say…. We’re getting hung up too much on this.
Sorry is I didn’t understand what your talking about and I’m sorry your not a consumer (you should be and I’m not saying that with any unpleasantness) get a OpenDrc and give it a spin.. just ignore all your points for awhile , get some fir filtering , make it the best you can without fir , then listen , then linearize the phase… you’ll see… it’s fun and rewarding.
Of course any system will be riddled with minutia of issues…. So what tho…. That doesn’t make it incorrect or not fun….
All the other non linearities are different subjects ….. I think that’s all I was trying to say…
I’m not being judgmental, I’m very much learning this also.
I understand your points, and that’s a different subject all together. That’s all I was trying to say…. We’re getting hung up too much on this.
Sorry is I didn’t understand what your talking about and I’m sorry your not a consumer (you should be and I’m not saying that with any unpleasantness) get a OpenDrc and give it a spin.. just ignore all your points for awhile , get some fir filtering , make it the best you can without fir , then listen , then linearize the phase… you’ll see… it’s fun and rewarding.
Of course any system will be riddled with minutia of issues…. So what tho…. That doesn’t make it incorrect or not fun….
I probably mix things up here, which is why I ask: assuming a speaker that is not running LR24 filters, but a asymmetrical crossover instead, could global phase linearization still be applied?
Exactly right,
Yeah the sum of everything no matter what angle each driver is at will be what is heard in the sums
Doesn’t mean your crossovers aren’t riddled with cancellations from misalignments, doesn’t mean they are either, that’s just going deeper into the picture…
You can tune to a target response and correct any phase and make it sound proper, it could be working speakers pretty hard if its bold negligence in crossover design, but you always make it sound good….
What’s really cool if getting the crossovers work really good first and keep the system efficiency up high…. Then do phase correction….
Yeah the sum of everything no matter what angle each driver is at will be what is heard in the sums
Doesn’t mean your crossovers aren’t riddled with cancellations from misalignments, doesn’t mean they are either, that’s just going deeper into the picture…
You can tune to a target response and correct any phase and make it sound proper, it could be working speakers pretty hard if its bold negligence in crossover design, but you always make it sound good….
What’s really cool if getting the crossovers work really good first and keep the system efficiency up high…. Then do phase correction….
Asymmetric crossover slopes can be advantageous to optimize in-room response, it's not bad design. If phase linearization is done at each way, linear phase crossover slopes are usually required and I was just not sure if this translates to global phase linearization. But it obviously does not, which is good news for my cheapo setup. With a 100€ CamillaDSP enabled RaspberryPi-streamer sourve into a Hypex plate amp, SOTA monitor speaker approaches can be copied, despite the Hypex are limited to IIR filters.
I have a question concerning a measurement result pre-FIR, IIR Hypex: why is the (minimum) phase of the waveguideed compression driver changing towards the very HF?
I have a question concerning a measurement result pre-FIR, IIR Hypex: why is the (minimum) phase of the waveguideed compression driver changing towards the very HF?
Looks like constant delay is a wee bit off, if zero degree phase is desired referenced to the highest frequency response (which I believe is best)I have a question concerning a measurement result pre-FIR, IIR Hypex: why is the (minimum) phase of the waveguideed compression driver changing towards the very HF?
Try adding a sample or so of delay to the measurement, using "Offset t=0"....on the main SPL tab or Impulse tab
Sure, give it a try.
Some folks seem to get great results, but most reports I've seen are very mixed.
Which makes sense i think, because although anything can be corrected near perfectly to a spot, the cost from using global phase correction comes from fixing things unique to that spot. (in addition to correction valid over a wider area).
Trick to getting good global results ime, is no doubt correcting a truly good global measurement.
@mark100
Do you think maybe it’s the vector average causing poor results maybe in smaller sized rooms especially?
The irratic responce differences between a vector average and let’s say a rms or db average , if that was corrected also and not just the phase could definitely lead to some pretty bad sound…
I seem to always get horrable dips where there are no dips, and the vector average is only useful for phase…
And I have another question….
Would a db average of 9 measurements around a head , and a generated minimum phase be a valid phase to use as a correction?
It’s only a minimum phase version….. but wouldn’t that also do the trick ?
Do you think maybe it’s the vector average causing poor results maybe in smaller sized rooms especially?
The irratic responce differences between a vector average and let’s say a rms or db average , if that was corrected also and not just the phase could definitely lead to some pretty bad sound…
I seem to always get horrable dips where there are no dips, and the vector average is only useful for phase…
And I have another question….
Would a db average of 9 measurements around a head , and a generated minimum phase be a valid phase to use as a correction?
It’s only a minimum phase version….. but wouldn’t that also do the trick ?