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

1ch in, in 1ch out. All measurement points can only be correlated to single input, a design axis is chosen that is representational of design application .....

Design point is 9" from tweeter face. Tweeter axis is angled about 5 degrees up from horizontal, and microphone is placed slightly below tweeter axis.
Presumably you have some clever method to put both your ears at that point. :D

Thanks for this Barley. The off axis vertical impulses show loadsa pre-impulse as expected and discussed by Greenfield.

You'll excuse me my old fashioned prejudice as a commercial speaker designer for more than half my life, that finds pre-impulse evil except for certain anti-aliasing filters.

I'm rather surprised at the amount of deterioration on the horizontal plane until I remember Pluto uses a 2" treble.

IMHO, Pluto only works cos the 2" treble and 1kHz xover. Of course a speaker is always the sum of the parts rather than dependent on any single spec of any single component so I'm sure it sounds OK.

You are perfectly right about xover freq. & driver separation. Pluto is the equivalent of my 'zillion dB/8ve' xover for subs at 80Hz.

The problem for a commercial speaker is finding a 2" unit with sufficient power handling. Most 'music' divides its power equally around 1kHz. Which is why its a good frequency for Bi-amping bla bla. I'll shut up about the evils of 2" directivity. :)

Aha! So you might know something about the so-called Stochastic Interleave alignment described here:
http://www.essex.ac.uk/csee/research...ssing LS.pdf
One of the papers I was external examiner at Malcolm's request is Rimmel, his ref 12.

Sadly, although Malcolm says the importance of his supa dupa method is "reducing the subjective significance of polar response errors", he doesn't show any evidence of the effect on polar response.

I think Greenfield's paper is a better discussion of this though I don't agree with his solutions.

Surprisingly, Malcolm also shows some very naive digital EQ, eg fig 2b, which Richard Greenfield and I have discussed many times and agreed as verboten!

Malcolm's paper, after you wade through the double integral obfuscation, is mainly about how to pretty up CDS curves by some very dubious methods, eg hiding the excess phase stuff. :eek:

The short answer is I dunno whether it "reduces the significance of polar response errors" and I don't think Malcolm knows either.

BTW, most of the Essex stuff was done with units provided by Celestion before I started work for them.

Re-reading "Is Linear Phase Worthwhile?", I can't fault any of my recommendations from 30 yrs ago even with all dis evil 21st digital stuff.
 
an interesting tool in rePhase is the All-pass inverse/All-pass function.

1: generate a inverse All-pass function to remove time,export...
2:generate a All-pass to add delay.export...
3:Import IRs with HOLM do the C=AxB (manual convolution),export as an impulse.wav

check group delay with REW.
it's useful in front of passive filter and a stereo amp,it avoids multi amp channel to delay some driver.
in this case,only middle band is delayed.
of course some FIR EQ are welcome with the time correction IR,no more cost !:p

An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
Administrator
Joined 2004
Paid Member
rePhasing a BR box

Happy New Year to all!
I've decided to enter a new phase for this year (sorry, couldn't resist)
Been working with rePhase and my woofers. On the chart below you will see 3 plots.
  • RED= Altec 416A woofer in a BR box measured at 5 feet (no crossover).
  • BLUE= Ideal 37Hz, 4th order, Butterworth high pass filter. Basically what the box is doing.
  • GREEN= Woofer and box response rePhased.
What will that sound like? Will I even hear it? Time will tell.
 

Attachments

  • BR-rePhased.png
    BR-rePhased.png
    22.8 KB · Views: 418
Administrator
Joined 2004
Paid Member
I am listening ONLY to the woofers. No crossover. As you can see, it's more or less flat up to 3.7Khz. I want to hear if I can detect any differences just in this very simple way.
So far, yes, I can hear it. It's subtle and not what I expected.

Adding in the mids and highs will be a completely different experience. As we know, our perception of one regiister (low, mid, high) is affected by the others.
 
Administrator
Joined 2004
Paid Member
I don't know John. I just exported the impulse wave from rePhase and plugged it into my player software - JRiver Media Center. I could be that it's upside down, I don't know. Will try to figure it out. I either have convolution turned on or turned off, if that's what you mean.
 
by default rephase linearization is an inverse all-pass.
the IR exported is corrected time

Pano,the HOLM graph you've shown is with a reverse time IR applied to correction,only to match and to check by visual.i think you've used a reverse option (in rePhase or in HOLM).

check your IR with REW,import IR as WAV,and choose "group delay" tab.
time in the lows should be negative.
 
So you do think this impulse when convolved will correct phase only of my woofer response?
That seems to be what I'm seeing, but could be wrong.

you're right,only in the lows and for the bass reflex delay.(picture below )
a simple way is to do a 4 feets mic measurement of loudspeaker,extract excess delay,and applied theoritical phase linearizationn for full range.

here's your correction (bass reflex 37 Hz)
An externally hosted image should be here but it was not working when we last tested it.
 
I don't know John. I just exported the impulse wave from rePhase and plugged it into my player software - JRiver Media Center. I could be that it's upside down, I don't know. Will try to figure it out. I either have convolution turned on or turned off, if that's what you mean.

If you turn the convolution turned on and off I don't think it's a valid comparison. What you need to do is keep the convolution on all the time and change the impluse being processed.