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

[TOINO]

Merci beaucoup, pos.

 

[natehansen66]

Cool, thanks!

 

[Pano]

Looks like some nice changes, thanks.

 

[pos]

version 0.9.8c is now online with a last minute bug correction

 

[davewantsmoore]

This is excellent. Thankyou Pos!

 

[pos]

rePhase 0.9.9 is online: rePhase | SourceForge.net

0.9.9 2014-12-10
    - shelving EQs with variable Q in Paragraphic EQ tab, with associated
      monotonic high and low shelv presets
    - centering can now be manually set to values in samples, percentage,
      time (us/ms/s) and distance (mm/cm/m).
      It is also possible to add or substract several values, for example
      "middle+270us"
    - new centering adjustment layout:
        * 'float' is now 'use closest perfect impulse' and is explicitely
          recommanded
        * 'int' is now 'round to closest sample'
        * 'use exact centring value' has been added for exact delays
    - 32 bits IEEE-754 float WAV output format added
    - output format noise floor is now shown in result curve
    - improved impulse and windowing symmetry, especially when an odd taps
      value is used
    - import/clear measurement file menu entries
    - fix partial installation catch

 

[orangeart]

Only Just noticed this Pos, a nice little Xmas pressie, thanks for your continuing hard work on this excellent app. It's already so good I can't imaging what you'll bring us in the big 1.0 release!

Stefan

 

[pos]

Thanks Stefan
Now you are putting pressure on me for the next release
Any suggestion for new features?

 

[orangeart]

Ha ha, sorry about that, please don't feel any pressure, it's frankly astounding that not only is this piece of software available for free but that it's available at all.

Whatever features you release in the future I'm sure will be most handy. I've got some 3 way speakers in development at the moments that I have specifically designed around FIR filtering and the flexibility it brings. Not loud, but low and tight using methods that just weren't available easily before.

Stefan

 

[micfre]

Any suggestion for new features?

Hi pos,
one thing that would help in generation of Horbach-Keele crossovers would be something to assist in calculation of the ratio parameter. I worked up something in Excel, but I am not super confident I got it right! (Pic and zip attached)

Cheers,
-m

 

[pos]

I am glad someone is actually using the Horbach-Keele feature
Adding the calculation directly in the main interface would be difficult, but I planned to add a "tools" menu entry with useful tools (left/right IR merging in a stereo file, offline batch convolution, passive filter calculation, unit conversions, etc.), and such a tool would certainly qualify.

Your zip looks empty tho

 

[orangeart]

I've never really looked into the Horbach-Keele filter slope before. I get that it can help with controlling directivity across the crossover region of two or more drivers although I'm not sure how. I can't see how comb filtering could be avoided.

Let's try a real world scenario, I've got a 3 way speaker where like most speakers the centre to centre distance between the mid and the tweeter is more than the ideal, obviously this causes some lobing and although I've not measured polarity I guess this looks worse as we move away from 0 deg. I've been using quite steep filters to keep the problem area to a minimum but am definitely interested in learning a different technique if these problems can be completely avoided?

micfre there doesn't appear to be a file inside you zip, maybe you could upload it again?

Stefan

 

[Crumboo]

orangeart: The Horbach-Keele filters are intended for symmetric loudspeaker arrays (e.g. WMTMW configuration) and gives a controlled vertical dispersion througout the frequency range. I've done some experimenting and like it. Here's a link to the original paper: http://www.xlrtechs.com/dbkeele.com...ear Phase Digital Crossover Flters Part 2.pdf

And +1 for a parameter calculation tool in rePhase!

Also, on the wish-list for next version is a tool for calculating cross-talk cancellation filters. Perhaps according to this paper: http://www.princeton.edu/3D3A/Publications/BACCHPaperV4d.pdf

Best regards,
Mattias

 

[Kindhornman]

Without having to go through this entire thread could someone point me any help files on how to use this what looks like very nice piece of software? Is there a calculator or simulation of any LCR impedance compensation networks for cancelling impedance rise rather than a simple Zobel network that can be placed on the speaker terminals? Thank for any direction and that POS for doing this great work for the community.

 

[pos]

And +1 for a parameter calculation tool in rePhase!

Also, on the wish-list for next version is a tool for calculating cross-talk cancellation filters. Perhaps according to this paper: http://www.princeton.edu/3D3A/Publications/BACCHPaperV4d.pdf

Hi Mattias

I would have to dig that up, I am not familiar at all with this subject.
Is it doable at all with a simple convolution?
In any case it will probably not be for there 1.0.0 release

 

[pos]

Without having to go through this entire thread could someone point me any help files on how to use this what looks like very nice piece of software? Is there a calculator or simulation of any LCR impedance compensation networks for cancelling impedance rise rather than a simple Zobel network that can be placed on the speaker terminals? Thank for any direction and that POS for doing this great work for the community.

Hi Kindhornman

Nice to see you here
I am afraid there is no help file or documentation...

The closest thing to a "10min introduction" would be the miniDSP application note: rePhase FIR Tool

rePhase is purely aimed at generating a FIR response to be used with a convolution engine. The way you define your target amplitude and phase correction curves is entirely manual (possibly based on an existing measurement). rePhase does not know how to interpret impedance plots, or predict their effects.

 

[Kindhornman]

Thanks POS,
The software looks interesting, I've been kind of peripherally watching the thread but didn't spend enough time reading the thread to know all that was going on. I'll have to start at the beginning and read through it. Thank you for your reply and working on something like this.

Steven

 

[pos]

I am planning an overhaul of the windowing implementation for the next version, and would like to remove a few no-that-useful (in filter design context) window algorithms.
Dropping support of course means that presets using them will not work as intended in future rephase versions.
So I need to know if some of you are using these windows, and/or would like to see them stay for some reason.

So here goes the endangered windows list:

• complex (only useful for phase-only correction when centering is set to exact sample values, which is not recommended anyway)
• cosine
• lanczos
• triangular
• bartlett
• bartlett-hann
• flat top (that one had a broken implementation, especially in the last version. it is now fixed but does not appear that useful...)

 

[jtalden]

pos, I am using rePhase for a phase correction filter rather than XO filter or others.

I have used the Complex window for several of my setups, but haven't investigated the choices deeply. With your intension to drop Complex, I finally did a more extensive trial of the other options. As it turns out the best fit for my situation is the Bartlett-Hann window. I would like to see that one stay, but I can probably get acceptable results with one of the remaining windows if you do remove it.

Thanks for making available to us a very useful tool!

 

[bambadoo]

Possible to implement:
kaiser-bessel ?
Parks–McClellan ?

Just wondering.
BTW: Pretty novice in this area/subject Thanks for the great work.