Making your own house target and create the FIR filter equalization file
open rePhase
create with filters and EQ the house curve you need.(slopes,low frequency boost ,subsonic filter )
export the target with target.txt for holm
export the target with target.wav for REW.
open HOLM or REW
A slot import the target.txt you've created with rephase.
B slot import the measurement (right or left at the listenning spot ).
change offset to merge the two curves,it will avoid high positive corr.
C=A/B
it gives the global equalization.
the green curve (C) is the result of difference between measurement and target.
the limitation of positive correction.
C=A x B
create with rephase an IR with negative EQ to reduce the too positive ones
export as .WAV,and make a stereo file with an another measurement.
same work can be done with REW or HOLM.
open rePhase
create with filters and EQ the house curve you need.(slopes,low frequency boost ,subsonic filter )
An externally hosted image should be here but it was not working when we last tested it.
export the target with target.txt for holm
export the target with target.wav for REW.
open HOLM or REW
A slot import the target.txt you've created with rephase.
B slot import the measurement (right or left at the listenning spot ).
change offset to merge the two curves,it will avoid high positive corr.
C=A/B
it gives the global equalization.
An externally hosted image should be here but it was not working when we last tested it.
the green curve (C) is the result of difference between measurement and target.
the limitation of positive correction.
C=A x B
create with rephase an IR with negative EQ to reduce the too positive ones
An externally hosted image should be here but it was not working when we last tested it.
export as .WAV,and make a stereo file with an another measurement.
same work can be done with REW or HOLM.
Last edited:
correction in green,simulated with REW
next,i'll sown measurement in room with this method
next,i'll sown measurement in room with this method
An externally hosted image should be here but it was not working when we last tested it.
Have you used linear filters generated by rePhase in J River convolution?
I just got my speakers setup using linear phase crossovers generated in rePhase. Horrible clicks and pops. I upped the latency on my interface to 1024 samples from 512 and it helped a bit (sounded like the woofers were clipping at 512 but they were nowhere near that). Still, the clicks pops and dropouts make it unlistenable as is.
I just got my speakers setup using linear phase crossovers generated in rePhase. Horrible clicks and pops. I upped the latency on my interface to 1024 samples from 512 and it helped a bit (sounded like the woofers were clipping at 512 but they were nowhere near that). Still, the clicks pops and dropouts make it unlistenable as is.
not me,few people are running with rephase,jriver and config file for crossover.it's ok.(DRC and FIR crossover ).
using the same sampling frequency is very important,48 kHz seems to be a good choice.
frequency of the source and Impulse must be the same.(otherwise it gonna troncate with the kind of phenomem you've describe ).
i can try with j river jukebox (freeware ).i gonna check...
in rephase,how any taps (samples) did you set for generation ?
default is 8192 taps,you can check with lower value.
in rephase,when you generate,take a look at the bottom right,it gives the lenght of impulse in second (taps/sampling frequency)
with J.river,be careful if you're using DRC convolver (like SIR) with crossover convolver.
you have to reorder plugin to set DRC before crossover.
Last edited:
Thanks, I'm running JR's convolver and everything is at 48khz. It's working fine now. I've been doing a lot of audio stuff the last couple days on this PC getting these speakers setup and didn't do the obvious........reboot the machine 
It's all good now.
Thanks again for your in-depth tutorials.
It's all good now.Thanks again for your in-depth tutorials.
I'm not doing any DRC yet, I wanted to get this going before adding anything else. I'm only using rePhase to build the crossover filters, and I do eq with JR's parametric equalizer. I'm not real crazy about rePhase's eq abilities and it's easier to make changes on the fly using the PEQ in JR. Phase comes out nice and flat this way so I'm happy.
I'm still having some issues with the audio stuttering though. I like to use the loopback function in JR for Netflix, Spotify, etc. In order to get the loopback to work without pops and clicks (at least when I was doing my eq and crossovers exclusively in JR's PEQ) I have to set the buffer super low, about .05 seconds. I can get clean playback from my hard drive in JR doing convolution with a higher buffer (2.5 seconds), but the loopback suffers. The loopback function doesn't work all that well with a low buffer and convolution though either. Gotta look into that.
I'm still having some issues with the audio stuttering though. I like to use the loopback function in JR for Netflix, Spotify, etc. In order to get the loopback to work without pops and clicks (at least when I was doing my eq and crossovers exclusively in JR's PEQ) I have to set the buffer super low, about .05 seconds. I can get clean playback from my hard drive in JR doing convolution with a higher buffer (2.5 seconds), but the loopback suffers. The loopback function doesn't work all that well with a low buffer and convolution though either. Gotta look into that.
rephase EQ is intend to correct definitvely direct field measurement.once done,no reason to change it
also you can add bass boost and subsonic filter in the crossover.
J River eq is not linear phase EQ,but it does not affect sound .(i do not listen difference phase linear and minimum phase ).
did you try less taps for IR.
the longest one is define by the lower frequency ,slope and ripple level.
depend of your sound card.
after this, apply the same number of taps for all impulse,band pass,high pass.
you may need only 1000 taps.(21 ms)
did you increase partitions in the convolver,paralleling process reduce delay.
also you can add bass boost and subsonic filter in the crossover.
J River eq is not linear phase EQ,but it does not affect sound .(i do not listen difference phase linear and minimum phase
).did you try less taps for IR.
the longest one is define by the lower frequency ,slope and ripple level.
depend of your sound card.
after this, apply the same number of taps for all impulse,band pass,high pass.
you may need only 1000 taps.(21 ms)
did you increase partitions in the convolver,paralleling process reduce delay.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
I understand that......it's getting it done that is the problem. Typically I build my eq filters in REW and copy them to the parametric eq in J River. Measurements confirm that REW's simulated eq'd response follows the actual output fairly close. On the fly changes are easy to make. The same set of filters plugged into rePHase's eq don't produce anywhere near the same result. The A + B trick gets kind of close to the actual result, but gets tiresome when you're working with 20+ filters
. The eq in rePhase is a bit buggy as well, at least in the Gain Bank.J River says they're working on convolution latency handling for the loopback function. I've set a second zone for now with min phase filters to use the loopback, but I'm going to look into the number of taps for the filters.
I don't know anything about how these FIR filters are built so this is a learning experience. I'm playing around with different numbers of taps in rePhase, and I'm wondering what the threshold for quality is (I suppose that's subjective). I've got a 35hz hi-pass (250 low pass), so right away I understand that's going to require more than a hi-pass at 80 would. The default 8192 taps shows a textbook roll-off and a nice continued slope as far down as the graph will go. At 5000 taps I'm getting ripple and a greatly reduced slope for the low pass starting around -100. For the 35hz high pass the slope starts transitioning to a shallower one around -40 and at 10hz -52 has transitioned to flat (no slope).
Any "rules" on whats acceptable here? I'm thinking anything greater than -50 or so wouldn't be audible at all or have any negative effect........but then I'm just guessing and like I said I don't know anything about linear phase filters.
I understand that......it's getting it done that is the problem. Typically I build my eq filters in REW and copy them to the parametric eq in J River. Measurements confirm that REW's simulated eq'd response follows the actual output fairly close. On the fly changes are easy to make. The same set of filters plugged into rePHase's eq don't produce anywhere near the same result. The A + B trick gets kind of close to the actual result, but gets tiresome when you're working with 20+ filters
it's possible,i never test the eq bank,only use the PEQ.it's working fine.
just to get easy conscience,with FIR EQREW seems to be buggy too,when using import impulse and making A+B,or A/B,AxB...(normalization,impulse centering ? )
i've heard something about that,but it was for previous version,(i believe MC 16 ).now,last version must be fixed i think.
no rules...your ears,120 dB without ripple is a little overkill.you are right,50-60 dB won't be audible,it's up to each other.
i can use 65000 taps without issue,no loopback function is needed with virtual cable
Last edited:
Hello,
you can run the DRC (8000 taps stereo )+(6-8 channels 8000 taps crossover ) with a basic laptop.
modern laptop or old desktop are more powerfull than external audio hardware (dolby lake,dx46...).
i do not believe in DAC sound,whatever a correct soundcard do the job (at least 90-95 dB S/N)
mini DSP has a 85 dB ratio (measured ).
Ahh......that might explain some of my problems! I'll try using Holm instead next time.
try to work with .TXT files only (impulse.txt,frequency.txt)
HOLM hates .wav files,it crashes every time with seven
about target curve.
https://docs.google.com/file/d/0B97zTRsdcJTfY2U4ODhiZmUtNDEyNC00ZDcyLWEzZTAtMGJiODQ1ZTUxMGQ4/edit?hl=en&pli=1
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
listening spot measurement at home,manual correction,7 points.(max 4 dB)
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
I understand that......it's getting it done that is the problem. Typically I build my eq filters in REW and copy them to the parametric eq in J River. Measurements confirm that REW's simulated eq'd response follows the actual output fairly close. On the fly changes are easy to make. The same set of filters plugged into rePHase's eq don't produce anywhere near the same result. The A + B trick gets kind of close to the actual result, but gets tiresome when you're working with 20+ filters
Hi
It is possible that the EQ implementation in rephase is different than the one in Jriver and REW.
There is a lot of different approches when it comes to Q interpretation in EQs (proportional, constant, etc.). Things get even worse with stuff like shelvings, where the corner frequency can be different from one implementation to another...
The EQ bank and paragraphic banks should give the exact same result, but it might be different to what Jriver, REW, or a DCX would produce.
I can try to improve this in a futur version, to mimic REW or another reference implementation more closely.
I will also probably implement an import for a measurement to be able to see the effect of the EQ (as well as filtering and phase tweaking) directly on measuremments instead of havving to use Holm or REW for the convolution.
Please do! I wouldn't change the way you implement the eq, but being able to see the effect in the same software would be huge.
The issues I'm having is with the Gain Bank. It only accepts + gains (by design?), and the numbers get stuck sometimes. If I mistype a value it gets "stuck" and won't change unless I clear the bank.
Other than those small issues (and my issues with J River which are being sorted) the software works quite well. I've always wanted to try linear phase but was put off by the cost of the software available so thank you for your efforts!
The initial intent with the bank EQ section was to be able to load EQs generated by another software (like REW), but that was never implemented (I wanted to stay away from automated corrections) and now I would like to remove these bank tabs altogether, and add some more paragraphic EQ tabs to replace them...
Regarding the negative gains entered with the numerical pad, yes it is a bug I will have to correct, thanks for the report!
In the meantime you can use the up and down keys to adjust values (this is the preferred method for most of the entires, like frequency, Q, slopes, gain, etc...), and that will give you negative gains (that you will also be able to correct with the numerical pad afterward).
Regarding the negative gains entered with the numerical pad, yes it is a bug I will have to correct, thanks for the report!
In the meantime you can use the up and down keys to adjust values (this is the preferred method for most of the entires, like frequency, Q, slopes, gain, etc...), and that will give you negative gains (that you will also be able to correct with the numerical pad afterward).
CPU load with 20 convolutions
cpu load with a basic laptop ( t4500 2 cores )
10 stereo convolutions of 8192 taps.(20 convolutions total )
the FIR crossover and FIR DRC can be done with Jriver (media center or jukebox (freeware)) and with any VST host
as we can see,any laptop is ok to do the job.
VSThost load 40%
cpu load 20 % (two cores ).
you can manage a 7.1 sytem (each tri ampified) for a total of 21 convolutions+DRC for each one with a laptop.
and a bass management (like velodyne sms 1 ) too.with high resolution in the low frequency (16000 taps ).
cpu load with a basic laptop ( t4500 2 cores )
10 stereo convolutions of 8192 taps.(20 convolutions total )
the FIR crossover and FIR DRC can be done with Jriver (media center or jukebox (freeware)) and with any VST host
as we can see,any laptop is ok to do the job.
VSThost load 40%
cpu load 20 % (two cores ).
you can manage a 7.1 sytem (each tri ampified) for a total of 21 convolutions+DRC for each one with a laptop.
and a bass management (like velodyne sms 1 ) too.with high resolution in the low frequency (16000 taps ).
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
same cpu (t4500 2 core laptop processor )
36 stereo convolutions of 8192 taps (72 total x 44100=26 billions ).
cpu load (VSThost use only 1 core )
91 %
36 stereo convolutions of 8192 taps (72 total x 44100=26 billions ).
cpu load (VSThost use only 1 core )
91 %
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.