rePhase, a loudspeaker phase linearization, EQ and FIR filtering tool

If you need more inspiration for flat sub bass in one point of the room I have an example
517565d1449133530-stm-32-f4-discovery-audioweaver-tried-eqbassboostphaseeqbrickwalllpfilter-jpg
I am not sure that is the best way to go making the phase flat when the amplitude response is not. I agree with wesayso that having the phase more closely follow generated minimum phase sounds better. What does generated minimum phase look like on that measurement?

Then calibrations are included/applied, I hope. Pos once found, I remember, that mic-calibration is not included when IR is exported from REW.
REW only applies mic calibrations to the SPL response but they are included when you export the measurement to text (not impulse).

I do not think I will linearize phase on sub bass level. The crossover is the only thing I will try to linearize, I think, for the speaker correction. But we'll see what happens :)
Try both linear and minimum phase crossovers and see if you can tell the difference, I find changing phase below is 500Hz is much easier to hear than changing it at higher frequencies.
 
Sorry, one final argument. This might be wrong? I have feeling the dip in the power responce at xover freq gets narrower, but deeper when higher order xover are used. And more linear power responce is prefered by most listeners. Again this might be utter noncence.

If my logic understanding of your expressed feelings is right, think you spot on power response dip at XO freq gets narrower the higher the order and think can be expressed as the logic that the higher the order the less the area can be distorted by adding a XO point, but its not any deeper than lower orders and here think the logic is they al sum at -6dB when talking LR topology. Think perfect linear power response is always good to dream about because real instruments and voices don't have this kind of distortion and it can be battled minimizing center to center distances, KEF do it perfect in a WT config using coaxial and most Synergy systems gets within 1/8 wavelength in a WTW config, we shall notice within same distances WT config always perform little better than WTW config.

Even it was a lot of work Rephase was once again very helpfull to study theoretical vertical power response reading slopes level numbers and set them into multiple instances of free XDir, to read them visuals focus on XO point is the one at left side in the middle and frq steps lower values up above and higher values down below, it doesn't matter 80Hz is used in example relative other XO frq will behave the same. First four graphs are LR 2nd/4th/8th/16th order within 1/4 wavelength and think notice they exactly the same at XO point but the higher the order improve on transition area, in last four graphs they all LR 4th order but first two are within 1/4 wavelength WT config then WTW config, and the last two are within 1/8 wavelength WT config then WTW config and think notice how less physical distances or combined relative lower XO points than mostly seen will improve power response using XO points. Of course graphs are all theoretical based and don't know anything about real world non ideal polars from used drivers or any physical nearby or room interference.
 

Attachments

  • LR2 Q-wave WT.PNG
    LR2 Q-wave WT.PNG
    669.6 KB · Views: 245
  • LR4 Q-wave WT.PNG
    LR4 Q-wave WT.PNG
    667.3 KB · Views: 244
  • LR8 Q-wave WT.PNG
    LR8 Q-wave WT.PNG
    665.7 KB · Views: 235
  • LR16 Q-wave WT.PNG
    LR16 Q-wave WT.PNG
    665.6 KB · Views: 237
  • LR4 Q-wave WT - Kopi.PNG
    LR4 Q-wave WT - Kopi.PNG
    667.3 KB · Views: 80
  • LR4 Q-wave WTW.PNG
    LR4 Q-wave WTW.PNG
    669.4 KB · Views: 86
  • LR4 O-wave WT.PNG
    LR4 O-wave WT.PNG
    666.9 KB · Views: 69
  • LR4 O-wave WTW.PNG
    LR4 O-wave WTW.PNG
    667.4 KB · Views: 69
Experience based on UMIK-1 / UMM-6 verse EMX-7150 / ECM8000 on MS OS is for me most responsive cleanest repeatable impulse response always works best when I/O is from same clock, though can't say if other users or OS used can maybe have other experience. Should decision be go analog way the calibration files are probably some notches more precise for few extra cost get either EMM-6 at CSL Cross·Spectrum Labs - Sound | Vibration | Engineering or ECM8000 at iSEMcon index.
 
Last edited:
Hey BYRTT, very nice work, thank you for sharing.

I've been playing with steep crossovers at 100 HZ for PA use...well I should say HiFi PA use...where the goal is killer sound quality first, SPL second.

Anyway, I think steep xover helps far more than it hurts.
It makes the critical region summation narrower, which helps with simple freq smoothing and phase matching thru that xover range, not to mention the directivity/ power response smoothing like you illustrated.

In small PA, you never know where the subs will be relative to mains. Sometimes mains sit on right on the subs in LR stacks , sometimes all the way separated with mains flown 20 ft in air.....and subs even in the center...you never know...

My experience is making the xover region as narrow as possible, minimizes the summation problem from all kinds of widely varying deployment geometries.
Get timing right, and with a steep xover it almost sounds like ventriloquism, with bass coming out of the mains despite the physical arrangement.

Same is true in room, IME.

I don't give a hoot anymore about pre-ringing concerns from steep filters using linear phase xovers.
In fact, I think steep filters probably have a bad reputation from IIR post ringing ! Or maybe better said, a bad rep from IIR highs' pre-ringing :D
 
@BYRTT @mark100

+1 Agreed and certainly has been my experience with time alignment and steep linear phase digital XO's. I have never worried about pre-ringing either. I have experimented with varying levels of pre-ringing and except for gross amounts, never hear it.

Some good info here in case folks have not seen it. Aside from a good overall read, search on "steep" in the text. Phase, Time and Distortion in Loudspeakers

PS. "ventriloquist" awesome! That's exactly what it sounds like. In small room acoustics, reducing/cancelling out the first major low frequency reflection can have this audible effect as well.
 
Ha ha mark100 you good there and impressive you also investigate and use modern filtering for PA touring, regarding posting about steep filters at this thread seems okay as long they minimum phase (IIR) but we shall expect some mumble in corners as soon linear phase (FIR) is on the table especially when combined steep slopes.

Can follow when you tour its challenge never know where subs will end up relative to mains but we can thank help from modern DSP there, well for fun i sometimes placed my woofer or mid-tweeter into neighbor territory distances just to hear how fun that sounds : )

You could be right about pre-ringing concerns, get timing right and don't do a bad repair, myself setup a 2-way 96dB/octave back in May/June and admit it was hard new territory get such two steep slopes feel its other at exactly same cycle but after that just enjoyed this bad looking exercise speaker, have since got dedicated computer hardware upgraded so JRiver DSP now runs flawless up at 192kHz with hundres of PEQs and also they released Jriver as real 64bit program, so with experience from past look forward setup same system with other woofer and a little higher frq for XO point so mid-tweeter get into a less stressing territory.

Pre-ringing and post-ringing :p when spacing being within 1/8 wavelength which one below will sound closets to real world sounds (measurement point about 74cm):
attachment.php
 

Attachments

  • 2006.png
    2006.png
    488.2 KB · Views: 382
Experience based on UMIK-1 / UMM-6 verse EMX-7150 / ECM8000 on MS OS is for me most responsive cleanest repeatable impulse response always works best when I/O is from same clock, though can't say if other users or OS used can maybe have other experience.
So just to confirm, in your experience you prefer the typical USB measurement microphones over the analog ones?
 
So just to confirm, in your experience you prefer the typical USB measurement microphones over the analog ones?

Sorry answer was not easy to decode meaning of, so good you asked : )

Prefer the analog microphones but reason is not because USB ones are not good, think its simply because they run their own clock domain not in sync with whatever soundcard output we use, and therefor in my setups using analog microphones bonded a interface using same clock always show most responsive cleanest repeatable impulse response compared to using USB microphones. Pretty shure a USB microphone records as nice as analog microphone but for record case signal flow is only into computer where when using measurement programs flow is supposed to be precision realtime I/O. For example often when using USB microphone IR/phase/distortion tabs into REW can look weird and will have to make re-sweeps up until one get something looking more normal and i don't blame USB microphone itself but the separate clocks because it never happen for analog microphones where I/O sits on same interface. But as said in previous post can not know if other users or OS used can maybe have other better experience. Another note could be should one down the road wan't to better read what happens above 20kHz area analog ones don't have the digital alias cut off and latest REW version have opened up non ASIO supported interfaces can now use Java drivers up at 88,2 plus 96kHz.
 
Last edited:
After installing the OpenDRC and checking the latency from our Antelope Orion 32 Spdif -> OpenDRC we measured 15ms latency after factory reset. With a FIR filter loaded it was 40ms. For reference we measured a DCX2496 with 100% processor load and it had a latency of 1,5ms.
15ms is too much for our application. With 40ms it's impossible to directly monitor and play piano or guitar. I'm returning the unit.
 
WalterPPK,

Thanks feedback, by the way for your build come to think about a couple minimum phase XO topology that give flat or relative flat phase.

"Filler" driver ala B&O

S. Harsch XO

First is very complicated in acoustic domain real world and demand with standard slopes a lot bandwidth of each driver, but there is link somewhere over there to spreadsheet from John Kreskovsky that can tune to other slopes than standard and output frd-files for those slopes.

Second think is much easier to dial in, its weakness can be in not being in 100% coherent phase in slopes transition area which will create a tilted lope as with 1st order filters, but if used in WTW config lobe direction on axis is repaired to strait forward and here think about your two woofers to midrange point.
 
Sorry answer was not easy to decode meaning of, so good you asked : )

Prefer the analog microphones but reason is not because USB ones are not good, think its simply because they run their own clock domain not in sync with whatever soundcard output we use, and therefor in my setups using analog microphones bonded a interface using same clock always show most responsive cleanest repeatable impulse response compared to using USB microphones.

This can still be a problem with PC DSP systems unless you are using a pro interface for a dac with a mic pre built in. In that case is it safe to assume the mic and dac are on the same clock? I use a pro interface although it does not have a mic pre. I have a separate analog pre and notice tenths of a ms shift between measurements when using a loopback timing reference because of the different clocks. This is workable but not entirely accurate.
 
This can still be a problem with PC DSP systems unless you are using a pro interface for a dac with a mic pre built in. In that case is it safe to assume the mic and dac are on the same clock? I use a pro interface although it does not have a mic pre. I have a separate analog pre and notice tenths of a ms shift between measurements when using a loopback timing reference because of the different clocks. This is workable but not entirely accurate.

If understanding right we probably close to same measurement setup then, myself use older M-Audio AP192 for I/O and separate analog mic pre, admit mostly stopped using loopback timing reference after reading some notes about how cross talk in bad layouts or bad user decisions regarding cable up system can leak and confuse reference, maybe I'm too suspicious here for measurement deviations but its not that hard for single driver sweeps do the alignment manual each time and for summing drivers there is other curves to look at to get them timed tight.

Thanks Byrtt for your help. I will read the threads.

Welcome, there had been much cleanup on www for free websites lately so should link to the brilliant spreadsheet from John Kreskovsky be broken send me a PM.