Interesting spelling of Voila when responding to somone with a French flag
Hello Australia !How was the new year night ?
you are in full summer.
Interesting spelling of Voila when responding to somone with a French flag
Lol , yeah that could be confusing 😹
"bad screen distance ** (251)"
I am getting this error repeatedly. It happens as soon as I try to import a file, and it also happens if I change the window size (via maximize/restore buttons) or if I try to manually resize the window. Nothing is ever displayed on the graph.
This happens on Windows 10 and Windows XP (both running in Virtualbox) and it also happens if I run rePhase in WINE.
I am getting this error repeatedly. It happens as soon as I try to import a file, and it also happens if I change the window size (via maximize/restore buttons) or if I try to manually resize the window. Nothing is ever displayed on the graph.
This happens on Windows 10 and Windows XP (both running in Virtualbox) and it also happens if I run rePhase in WINE.
I'm having a bit of trouble defining Horbach-Keele filters in rePhase v1.4.3.
I'm replicating the filter responses from the 5-way example system detailed in Part 2 of H&K's AES presentation. I've had luck doing this in the demo version of Fir Designer M, as well as VituixCAD. VCad's simulated polar responses seem to match those in the paper, which seems encouraging.
But in rePhase, defining an HP or LP with "Shape" "Horbach-Keele" in the Linear Phase Filters tab starts with a default "Param" of 2.5. This is H&K's "R" parameter for the filter.
I need some filters with R in the 3.x range.
Unfortunately, any attempt to edit the first digit in this edit box results in the value clamping to either 1.0 or 4.5. I suspect these are the minimum and maximum allowed. I've tried every combination of the Delete and backspace key I can think of, but darned if I can figure out how to get rid of that "2" and replace it with a 3.
The end result seems to be that with great care, one can create an H-K filter with R values between 1.00 and 2.99, or 4.00-4.50, but I can't figure out how to get anything between 3.00-3.99.
v1.4.2 seems to have the same behavior. Can anyone think of a workaround?
Thanks!
I'm replicating the filter responses from the 5-way example system detailed in Part 2 of H&K's AES presentation. I've had luck doing this in the demo version of Fir Designer M, as well as VituixCAD. VCad's simulated polar responses seem to match those in the paper, which seems encouraging.
But in rePhase, defining an HP or LP with "Shape" "Horbach-Keele" in the Linear Phase Filters tab starts with a default "Param" of 2.5. This is H&K's "R" parameter for the filter.
I need some filters with R in the 3.x range.
Unfortunately, any attempt to edit the first digit in this edit box results in the value clamping to either 1.0 or 4.5. I suspect these are the minimum and maximum allowed. I've tried every combination of the Delete and backspace key I can think of, but darned if I can figure out how to get rid of that "2" and replace it with a 3.
The end result seems to be that with great care, one can create an H-K filter with R values between 1.00 and 2.99, or 4.00-4.50, but I can't figure out how to get anything between 3.00-3.99.
v1.4.2 seems to have the same behavior. Can anyone think of a workaround?
Thanks!
I did an HK in a car on a 5 way …. 1024 taps on all drivers except sub and midbass which was done on a pair of OpenDrcs
1 OpenDrc for sub
1 for midbass HPF (which was inherent on all of the other speakers as all the mains ran through that OpenDrc also) and 2x4hds everywhere else.
Works killer! The super narrow bandwidth on the lower midrange worked way better then I expected. It took a lot of time to get right but was really cool
Do u have a way to run a live IR (smaart or systune) ?
The low taps demanded that I work the fir live and not from an imported measurement. The end result is stellar tho
Actually, it's a pile of Symetrix Edges. I have a pile of them, and would like to use a couple for this, if it passes initial muster. The project is a 5 way (horizontally) symmetric line array.
I'm hoping that the saving grace re; the tap limit is that the signal flow allowed by the Symmetrix Designer is pretty arbitrary.
There plenty of DSP headroom to either do a mixed IIR/FIR solution, or cascade a couple of 1024 tap FIR filters for the LF 'ways'. It seems like Symetrix could have provided a 2048 tap FIR block, but I'm guessing their target audience (including me, most of the time) wouldn't be able to cope with the latency.
The downer is unless I've been missing something, they don't offer any SRC conversion blocks, so the whole signal flow seems to be at 48K.
Anyway... I was pretty sure that with the Edges 'tapping out' (har har) at 1024, I was going to have to prototype this project using a PC-based convolution engine, and use Dante VSC or Dante VIA to send the processed results to the Symetrixes (Symmetrii ??). I still might do that.
But after loading the H-K paper's 5 way system's critical frequencies into FirDesigner and (now) rePhase, I was a bit surprised to see that the target curves for most of the filters can be matched pretty well with 1024 taps.
The subwoofer's low pass (R=2.5, critical f=119Hz), with middle centering and a cosine window, looks pretty OK. That kind of blew my mind.
The upper-mid bandpass (HPF H-K R = 3.0, f=825, LPF "HK-Last" F = 2475) and HF are pretty perfect. The lower-mid section is pretty OK, if I squint a bit.
The woofer bandpass (HPF R=2.5, F=119, LPF R=2.78, f= 297Hz) is the big problem. The amplitude on the HP side of the filter really misses the target starting below ca. 135Hz. The phase is OK in-band, but the amplitude can't track the target curve below 150Hz, and there is a lot of OOB signal below that.
Using something other than cosine windowing just seems to push the problem around...
I did a quick check with FIR Designer M's demo to see what dropping an IIR HP or a PEQ notch ahead of the FIR does... the amplitude and OOB behavior looks better, but the impulse, after the FIR compensated for the IIR filters' phase shift, started to look really asymmetric. Wouldn't that bode poorly for any hope of the woofers' FIR pre-ring getting cancelled out by the subwoofers' ?
Anyway, I'm open to suggestions as to how to improve things, if indeed it isn't mathematically impossible. Please let me know if screenshots, .rephase files, etc, would be useful.
I'm hoping that the saving grace re; the tap limit is that the signal flow allowed by the Symmetrix Designer is pretty arbitrary.
There plenty of DSP headroom to either do a mixed IIR/FIR solution, or cascade a couple of 1024 tap FIR filters for the LF 'ways'. It seems like Symetrix could have provided a 2048 tap FIR block, but I'm guessing their target audience (including me, most of the time) wouldn't be able to cope with the latency.
The downer is unless I've been missing something, they don't offer any SRC conversion blocks, so the whole signal flow seems to be at 48K.
Anyway... I was pretty sure that with the Edges 'tapping out' (har har) at 1024, I was going to have to prototype this project using a PC-based convolution engine, and use Dante VSC or Dante VIA to send the processed results to the Symetrixes (Symmetrii ??). I still might do that.
But after loading the H-K paper's 5 way system's critical frequencies into FirDesigner and (now) rePhase, I was a bit surprised to see that the target curves for most of the filters can be matched pretty well with 1024 taps.
The subwoofer's low pass (R=2.5, critical f=119Hz), with middle centering and a cosine window, looks pretty OK. That kind of blew my mind.
The upper-mid bandpass (HPF H-K R = 3.0, f=825, LPF "HK-Last" F = 2475) and HF are pretty perfect. The lower-mid section is pretty OK, if I squint a bit.
The woofer bandpass (HPF R=2.5, F=119, LPF R=2.78, f= 297Hz) is the big problem. The amplitude on the HP side of the filter really misses the target starting below ca. 135Hz. The phase is OK in-band, but the amplitude can't track the target curve below 150Hz, and there is a lot of OOB signal below that.
Using something other than cosine windowing just seems to push the problem around...
I did a quick check with FIR Designer M's demo to see what dropping an IIR HP or a PEQ notch ahead of the FIR does... the amplitude and OOB behavior looks better, but the impulse, after the FIR compensated for the IIR filters' phase shift, started to look really asymmetric. Wouldn't that bode poorly for any hope of the woofers' FIR pre-ring getting cancelled out by the subwoofers' ?
Anyway, I'm open to suggestions as to how to improve things, if indeed it isn't mathematically impossible. Please let me know if screenshots, .rephase files, etc, would be useful.
Sorry about that... I tried to edit the blank post, but took too long to do so. It's probably OT, but here it is:
No I don't have a realtime way to live-audition FIR filters, unless I pony up and buy FIR Designer M. Eclipse has recently that (and a/b preset comparisons) to "M", but I've resisted the purchase so far. It's a fairly expensive annual subscription. Paying x $$$ a year forever to be able to tweak my systems is a pill that I so far have resisted swallowing.
What I've been doing for mixed FIR/IIR crossovers is pretty much the opposite of a live workflow...
-REW dual channel raw driver measurements (typically nearfield-ish, sometimes outdoors)
-REW IIR min-phase driver EQ calculation to some target curve or another (flat in passband + 1 octave, or an IIR filter)
-manually load PEQ to Symmetrix
-remeasure corrected driver in REW
-Export measurement to rePhase
-Create/fiddle with FIR crossover
-Export filter as .csv
-Rename .csv to .txt so Symetrix Composer will read it (ugh).
-Load said filter to the hardware
-Remeasure & listen.
One 'way' at a time. It's pretty painful, especially to get an initial working configuration for the 4+ way systems I'm partial to.
My quick and dirty all-IIR Symetrix, DriveRack, etc, workflow is live, and is literally 200x faster...
-Use an OmniMic mounted on a long counterweighted studio boom
-Use the OmniMic app in continuous measurement mode
-Tweak each driver's EQ for flatness (live, on the hardware's app) one at a time, waving the mic about in x+y to get some averaging
-Un-mute top two drivers, enable pairs of crossovers, start tweaking delay, phase, order, Q, etc, until I like the phase & mag response I see.
-Wave the mic around some more to see if there are bad lobes near the design axes.
-Rinse and repeat for the lower 'ways', typically requiring some iterative tuning of the adjacent bands.
It is vastly less repeatable, documentable, etc, and totally unsuitable for FIR, but lets me try so many more variations in so little time that I often prefer the end results, at least for simpler systems. I do the whole thing outdoors (for PA), or in my fairly large (and very overdamped, at the speaker end) room. Indoors, I get immediate feedback when I'm doing something dumb; I can often hear it in room's response to the sweeps even before I start probing around with the mic, or switch from the stimulus to music.
I suspect that being able to hear tweaks to FIR filters live would be a very valuable learning experience, especially at least unless/until I gain a firmer understanding of what makes FIR filter sound 'bad', temporally, or otherwise.
No I don't have a realtime way to live-audition FIR filters, unless I pony up and buy FIR Designer M. Eclipse has recently that (and a/b preset comparisons) to "M", but I've resisted the purchase so far. It's a fairly expensive annual subscription. Paying x $$$ a year forever to be able to tweak my systems is a pill that I so far have resisted swallowing.
What I've been doing for mixed FIR/IIR crossovers is pretty much the opposite of a live workflow...
-REW dual channel raw driver measurements (typically nearfield-ish, sometimes outdoors)
-REW IIR min-phase driver EQ calculation to some target curve or another (flat in passband + 1 octave, or an IIR filter)
-manually load PEQ to Symmetrix
-remeasure corrected driver in REW
-Export measurement to rePhase
-Create/fiddle with FIR crossover
-Export filter as .csv
-Rename .csv to .txt so Symetrix Composer will read it (ugh).
-Load said filter to the hardware
-Remeasure & listen.
One 'way' at a time. It's pretty painful, especially to get an initial working configuration for the 4+ way systems I'm partial to.
My quick and dirty all-IIR Symetrix, DriveRack, etc, workflow is live, and is literally 200x faster...
-Use an OmniMic mounted on a long counterweighted studio boom
-Use the OmniMic app in continuous measurement mode
-Tweak each driver's EQ for flatness (live, on the hardware's app) one at a time, waving the mic about in x+y to get some averaging
-Un-mute top two drivers, enable pairs of crossovers, start tweaking delay, phase, order, Q, etc, until I like the phase & mag response I see.
-Wave the mic around some more to see if there are bad lobes near the design axes.
-Rinse and repeat for the lower 'ways', typically requiring some iterative tuning of the adjacent bands.
It is vastly less repeatable, documentable, etc, and totally unsuitable for FIR, but lets me try so many more variations in so little time that I often prefer the end results, at least for simpler systems. I do the whole thing outdoors (for PA), or in my fairly large (and very overdamped, at the speaker end) room. Indoors, I get immediate feedback when I'm doing something dumb; I can often hear it in room's response to the sweeps even before I start probing around with the mic, or switch from the stimulus to music.
I suspect that being able to hear tweaks to FIR filters live would be a very valuable learning experience, especially at least unless/until I gain a firmer understanding of what makes FIR filter sound 'bad', temporally, or otherwise.
Specific to trying to optimize the 119Hz H-K HFP that I'm having trouble 'fitting' into a 1024 tap FIR:
I'm finding that, at least on paper (or phosphor...), it is quite informative that in Designer M, I can interactively tweak an 'upstream' IIR filter while seeing if/how the tap-limited FIR's windowed mag & phase (and stop-band) response improves (or not!).
It would be really cool if REW + rePhase could someday talk to each other in nearer-to-realtime to achieve a similar result. Or can it?
If not, I'm wondering if it could be as simple as rePhase having a checkbox to monitor an imported measurement (or impulse) file for changes, and for REW to have an option to automatically update an export file in response to changes...
I'm finding that, at least on paper (or phosphor...), it is quite informative that in Designer M, I can interactively tweak an 'upstream' IIR filter while seeing if/how the tap-limited FIR's windowed mag & phase (and stop-band) response improves (or not!).
It would be really cool if REW + rePhase could someday talk to each other in nearer-to-realtime to achieve a similar result. Or can it?
If not, I'm wondering if it could be as simple as rePhase having a checkbox to monitor an imported measurement (or impulse) file for changes, and for REW to have an option to automatically update an export file in response to changes...
Smaart and FIR designer work together.
They have each other‘s software embedded so they talk to each other it’s super cool but I am a full-blown diehard RePhase fan. I’m super loyal once I get something that works and I absolutely love multi mic averages in smaart. It truly is the best way. I can get good phase to 20k (not kidding)…
The problem with REW is you can’t average the phase. So a single mic is only good to like 1k ish … 2k if your good…
I basically align all my speakers with a single static mic time using the tweeter as the mic time…. Then align with dsp delay all drivers to impulse peak single mic
Then I bring out the big guns and do a mic array around the head space and get a multi mic average. The phase is pristine the way smaart removes the time of flight… it’s perfect. The dual ch is a must for FIRs IMO.
I did not know that REW had dual ch… I know there was some chatter about it on the AVNirvana pages , but I never saw anything happen with it..
I have the pro version of REW and it works… it’s nothing like smaart.
I’ve heard of people getting pirate versions of V8…. I don’t know if you like to wear a patch over your eye and have a parrot on your shoulder… food for thought.
But it sounds like what you’re doing is pretty cool. And if you were getting results that you like then, there’s no arguing with results.
They have each other‘s software embedded so they talk to each other it’s super cool but I am a full-blown diehard RePhase fan. I’m super loyal once I get something that works and I absolutely love multi mic averages in smaart. It truly is the best way. I can get good phase to 20k (not kidding)…
The problem with REW is you can’t average the phase. So a single mic is only good to like 1k ish … 2k if your good…
I basically align all my speakers with a single static mic time using the tweeter as the mic time…. Then align with dsp delay all drivers to impulse peak single mic
Then I bring out the big guns and do a mic array around the head space and get a multi mic average. The phase is pristine the way smaart removes the time of flight… it’s perfect. The dual ch is a must for FIRs IMO.
I did not know that REW had dual ch… I know there was some chatter about it on the AVNirvana pages , but I never saw anything happen with it..
I have the pro version of REW and it works… it’s nothing like smaart.
I’ve heard of people getting pirate versions of V8…. I don’t know if you like to wear a patch over your eye and have a parrot on your shoulder…
food for thought.But it sounds like what you’re doing is pretty cool. And if you were getting results that you like then, there’s no arguing with results.