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

...Can somebody help with the rephase export settings for targetting JRiver's convolver?...

About format all of them in red below should work, in past using the 32bit JRiver version myself used 32 bits IEEE and after they released 64bit JRiver version i started use 64bit IEEE, but again all of the red ones would probably work inside their 64bit engine. Think all the other settings should more or less say itself, but can recommend if lag is no problem that huge taps and FFT lenghts combined "use exact centering value"/"rectangular" window sounds extra great and natural to my ears, but know not all users can live with huge lag of sound stream.

If JRiver automatic shifts rate (dynamic/bitperfect settings) from track to track one can ignore that and simply point to whatever wav-file with a user set rate and JRiver will resample filter for that single convolution file, but if one wan't correctly dynamic sample rate switching there is guides below:

Convolution - JRiverWiki

Convolution Sample Rate Switching

attachment.php
 

Attachments

  • 1000.png
    1000.png
    48.4 KB · Views: 930
There is no plan for auto EQ, but you can import auto EQs generated with REW

I have a somewhat strange question (as a rePhase rookie). I have been using Equalizer APO to DSP my speakers for some time now, I have only recently started experimenting with REW and rePhase.

However, after auto EQing with REW, I dont like the sound of my system. Generating a theoretically ideal phase-linear crossover at the desired crossover points with rePhase sounds MUCH better as opposed to measuring each driver, linearizing, EQing and then applying the crossover in rePhase.

Im wondering, why that might be? I assume my measurements are garbage and thus the new perfectly linear system is in fact not so linear. It sounds muffled.

Do you have any advice or what is generally the best method to achieve a good crossover?

Assume I have a two-way speaker. Would it be preferably to
1. measure each driver, EQ and linearize in REW, then create a crossover in rePhase for each driver.
2. create an ideal crossover in rePhase. Measure the whole system in room and linearize everything in REW and apply a correction filter in rePhase?

Thanks for this great piece of software!
 
Hello hugligan

Measurements play the most important part here: the way you take them, analyse them and process/prepare them is very important, especially when using auto EQ.
That is why many automated solution also include a very specific measurement procedure.

There is no ideal electrical crossover when it comes to real world drivers: you need to get a precise idea of the driver's response (in band, but also in the stop bands) to get the acoustical response you are looking for. Getting the acoustical crossovers right for both magnitude and phase is the most important part.

Here is an example the procedure I personally use: https://www.diyaudio.com/forums/mul...ion-eq-fir-filtering-tool-68.html#post4322701
You can of course choose to use automated EQs instead of manual EQs as you see fit.
 
I have a somewhat strange question (as a rePhase rookie). I have been using Equalizer APO to DSP my speakers for some time now, I have only recently started experimenting with REW and rePhase.

However, after auto EQing with REW, I dont like the sound of my system. Generating a theoretically ideal phase-linear crossover at the desired crossover points with rePhase sounds MUCH better as opposed to measuring each driver, linearizing, EQing and then applying the crossover in rePhase.

Im wondering, why that might be? I assume my measurements are garbage and thus the new perfectly linear system is in fact not so linear. It sounds muffled.

Do you have any advice or what is generally the best method to achieve a good crossover?

Assume I have a two-way speaker. Would it be preferably to
1. measure each driver, EQ and linearize in REW, then create a crossover in rePhase for each driver.
2. create an ideal crossover in rePhase. Measure the whole system in room and linearize everything in REW and apply a correction filter in rePhase?

Thanks for this great piece of software!


Also remember to use time alignment offsets on each drivers fir.
And add/subtract the correct values .

I like to also when I’m done making crossovers with pos method listen and ajust the delay a few clicks +/- about .02- .5ms each way to listen for a better alignment , sorta depends of the frequency I’m trying to align for delay time (in case my measurements weren’t perfect) sometimes a tiny bit of delay puts it right where it needs to be as a sum. It’s amazing how quickly a bad alignment it becomes as messy. You’ll hear it and than measurements will / should validate
 
Last edited:
Hi all, Just started experimenting with REW and Rephase. I'm having some trouble importing the settings from REW to rephase though. Both with the latest version (1.4.3) and the previous version (1.4.2) i get the error message that the measurement file is in the wrong format.

I've just exported it out of REW from the eq window choosing Rephase as formatting style. Does anyone know what is going wrong here?
 
It looks like you are trying to import EQ settings as measurements, which they are not.
You need to go to the "Paragraphic Gain EQ" tab, choose a bank, and use the "import REW filter settings" function in the "Tools" drop-down menu.
Keep in mind that every existing EQ points set in this bank will be lost when importing REW settings.
 
Good day!
First of all I want to thank you for your hard work! Great program! Thanks. I'm trying to figure it out. I am not a physicist and it is not easy for me :) How to turn the phase to 180? I have a task at frequencies of 100-150 Hz through the FIR filter to expand the phase. How can I do that ? Thank!
 
Hello grizzli,
Thanks for the kind words :)
Turning the phase 180° for the whole frequency range is simply done by reversing polarity, but I guess this is not what you want to do here.
Altering phase in a restricted frequency range can be done in rephase using the filter linearization tab, or the phase EQ tab.
I would advice you to preferably rely on the filter linearization tab, and only correct existing phase shifts, and not necessarily everything that a given measurement show you. It is very easy to be fooled by a measurement and over-correct things that should be left alone such as comb filtering from reflections, diffractions, etc.
 
Hello grizzli,
Thanks for the kind words :)
Turning the phase 180° for the whole frequency range is simply done by reversing polarity, but I guess this is not what you want to do here.
Altering phase in a restricted frequency range can be done in rephase using the filter linearization tab, or the phase EQ tab.
I would advice you to preferably rely on the filter linearization tab, and only correct existing phase shifts, and not necessarily everything that a given measurement show you. It is very easy to be fooled by a measurement and over-correct things that should be left alone such as comb filtering from reflections, diffractions, etc.

I do not understand. Unable to deploy phase as you want. Something I am doing wrong.
I have to settle like this. Is it possible
 

Attachments

  • 1.JPG
    1.JPG
    131.3 KB · Views: 394
  • 2.JPG
    2.JPG
    188.5 KB · Views: 391
Is it a 100 to 150Hz measurement?
I am afraid I cannot provide any meaningful advice based on these screenshots.

Yes, I just need to turn these frequencies. Measurement made them. The full range of frequencies does not work, some kind of incomprehensible schedule comes out. I can throw the result of measurement.
I would be very grateful if you look.
 

Attachments

  • Measure.txt
    3.6 KB · Views: 47
Yes, I just need to turn these frequencies. Measurement made them. The full range of frequencies does not work, some kind of incomprehensible schedule comes out. I can throw the result of measurement.
I would be very grateful if you look.
If you want an accurate measurement of that range you need to start the sweep lower in frequency.
Moreover, you also need to see what is going on around that range to get a good assessment of the situation.

Regarding that particular measurement, the passband of its magnitude response appears to be around 80dB, so you need to set the gain offset to around -80dB in the measurement tab to see it.

Using minimum-phase EQ to "flatten" the magnitude response (quotes here to emphasis the fact that attempting to flatten anything in such a tiny range is not really meaningful) will already tend to flatten the phase response.
 
Hi Thomas


Time to waste ?
a constant slope from a given frequency ?
it can be done with low Q EQ but not really flat.
:) i'm getting out...


190715024446753353.png


190715030052488857.png

Hi Thierry,

I am afraid such a constant slope would not be physically "sustainable" :D
You ought to have a knee somewhere otherwise the magnitude would be infinite at DC, no matter how gentle the slope.
If you are looking for a linear-phase response then using a LR filter with a low slope (can be set as low as 1dB/oct manually) would work. A "reject low" low-pass filter would also work, with a more angular knee.
For minimum-phase response a shelving filer (in the EQ tab) with a very low Q would work.
 
Re-Hi Thomas,
Linear phase or min. phase do not allows this "soft" slope from a given frequency (rePhase 1.4.3)
No needs to start from DC,but something like about ~500/2KHz.



neither with LR choice (min 12dB/oct) nor low shelving..
This is not important i can do it with min.phase EQ.:eek:


Just a suggestion.


Edit:with a linear phase tab,it'ok (but it's starting at "DC")
no frequency choice for "cutting"
 
Last edited: