Too bad...This would be ok in the final step, when the user defines his requirements, and acepts or not sacrifying quality or conveniente, more or less taps, and the red curve informs how good or bad the result will be, not before. Butt18, or whatever, is Butt18. Will have to check twice before using rephase...
Hi, the only non-minisharc FIR capable minidsp I know of, is the hd 2x4
.....but check the manual....
Last edited:
Hi GDO, not "too bad"....been using 6144 taps. This method gives good red curve at half that...
It looks like the problem is that the blue curve (ideal one) is not reliable because of also tap number dependent. It sounds like an awfull mess...
I am comfortable with the decisions I took with those subsonics, but I agree that it should be documented, and non "optimized" options should also be proposed. You should realize though that a 90° shift at DC is not something that can be done, so it would get approximated in the end anyway, gobbling taps like crazy for an impossible target and potentially damaging other correction settings on its way.
These linearization filters settings are restricted to subsonics only, and are not supposed to be use for any other purpose (only LR linearization is proposed for crossover, which is the only logical way to operate). The approximations take place under fc, which with subsonic will generally be under 25Hz, and with an attenuated magnitude signal...
Anyway, do as you please
Last edited:
Hi charlie2
As mark100 noted only the openDRC, miniSharc and 2x4HD do have an actual FIR section.
That said, using the advanced version of the minidsp plugin you should able to build relatively high order minimum-phase crossover by using many biquads, each one potentially adding a 12dB/oct slope.
You can look for Butterworth tables that will give the Q coefficients to use to get a given target Butt order, and then use 2 Butt to get an LR (and of course as always the target is the actual acoustical filter, not the electrical one).
In the end the resulting phase shift might be linearized using a computer or fir-capable hardware on the stereo input signal.
GDéon said:
Maybe a future Nobel price.
Taking analog circuit,what solution is available with DC values ?
Ex:an OP amp with k factor amplification.
solution 1,output DC value have the same sign.
solution 2,output DC value is inverted (180° phase shift,or j²=-1).
Between this,(DC value) no intermediate phase can exist.
No more way with FIR/IIR .
or your subsonic filter is not high-pass.(zeros on numerator).
DC is a theoretical limit i don't care about: I am not the programmer!
Btw, i understand a limit must be put to both max tap number, lowest practical frequency limit at a given sampling freq. Just decide and inform about this is part of programmer's mission.
In other words, if i am ready to accept the latency of a 131k taps fir at 44100 for a 10, 15 or 20hz cutt off, or whatever, it's my problem, not the programer's business...
And once again, if in Rephase , the blue curve is the ideal one, and the red curve the compromised one, why compromise with the blue one?
Last edited:
GDO, your remark is duly noted and I will make sure I add a notice in the preset name for the next version (something like "Butt XXdB/oct optimized approximation"), as well as normal versions for the die hard subsonic fans
Next version should be out soon as the bug jmbee reported (2nd order MP filters with Q<.5) is a major one and has to be corrected ASAP.
It is a little bit earlier than I wanted, although it has already been 15 months since the last update
Next version should be out soon as the bug jmbee reported (2nd order MP filters with Q<.5) is a major one and has to be corrected ASAP.
It is a little bit earlier than I wanted, although it has already been 15 months since the last update
Is there software that will apply a convolution on top of another one?
I have a use case in mind where I use one piece of software ( FIR Designer ) to idealize driver response, and use rePhase to apply a linear phase crossover on top of its convolved output.
Of course I can manually tweak rePhase to achieve excellent results, and that is what I do today. I do like, however, the power of FIR Designer's measurement averaging and automated magnitude and phase adjustment. rePhase offers much more flexible and suitable (for my setup) crossover options.
Thanks for reading.
I have a use case in mind where I use one piece of software ( FIR Designer ) to idealize driver response, and use rePhase to apply a linear phase crossover on top of its convolved output.
Of course I can manually tweak rePhase to achieve excellent results, and that is what I do today. I do like, however, the power of FIR Designer's measurement averaging and automated magnitude and phase adjustment. rePhase offers much more flexible and suitable (for my setup) crossover options.
Thanks for reading.
Last edited: