Hello, I am new to rephase - a beginner.
I would simply like to use graphic EQ and phase EQ.
I have a win10 laptop as soundsource.
I can measure my one way fullrange loudspeakers with arta.
So is there an easy way just to play the music from the laptop, use the filters from rephase and to listen and check phase and frequency response with arta?
I want to build backloaded horns and want to see if the time fault generated by the horn can be corrected.
I would simply like to use graphic EQ and phase EQ.
I have a win10 laptop as soundsource.
I can measure my one way fullrange loudspeakers with arta.
So is there an easy way just to play the music from the laptop, use the filters from rephase and to listen and check phase and frequency response with arta?
I want to build backloaded horns and want to see if the time fault generated by the horn can be corrected.
Last edited:
@Freedom666
So , unfortunately it will be hard to validate phase after unless your mic positions are exactly the same….
Even with a dual ch analyzer it’s hard…. Phase delay will always make its way I to the measurements….
If you use a single mic position you could validate some LF phase up to about 300hz , but much above that phase delay will always get in your measurements… a little bit.
Up to about 1k you can get pretty good phase validation with a single mic position and a dual ch analysis… REW won’t cut it .
REW is great for getting good measurements and then making a average and setting phase that way or close to speaker measurements.
With REW you need to do an average of many measurements around your head and the measurements must be coherent (vector averages). The issue with the vector average is that the frequency response has serious problems…
So you need to do a 1st session using db averages, or spl averages to do eq work, then the vector average for phase correct only.
I would check out open sound meter it’s a free dual ch analyzer that’s free…. You’ll need an XLR mic and Interface that has a loop back…. Then you can take some Polar averages and get great phase data and great frequency response data , capture the average around your head (9 second temporal average works nice) make impulse , send to REW , then export as text to rephase for work
An fir does exactly what you ask, if you get in a situation where your fir didn’t do what it was supposed to that means an acoustical issue that will always have that phase signature no matter what you do… so again, validation is tough….. but can be done if you understand the nature of mics and phase and phase delay and making sense of “good enough “
Whatever you capture for left and right ch , and change the ir you sent to rephase should be valid within it self ….. as any anomalies will be fixed for that mic position, or it’s average …. It will just be extremely hard to get that exact same sequence unless you have mic that is fixed in one spot while you go directly to validation stage, but then it’s only valid at one tiny spot. So you must make an average of many spots…. So catch 22 , you’ll have a hard time validation but again , read between the lines. Think logically, and don’t ever assume the fir didn’t do what you asked , instead look at why it isn’t showing what you changed and look at your room or setup
Programs like firD can talk directly to smaart and do a validation live as it makes an iterative fir session for you ….
Best of luck , read this thread for more info it’s all there
So , unfortunately it will be hard to validate phase after unless your mic positions are exactly the same….
Even with a dual ch analyzer it’s hard…. Phase delay will always make its way I to the measurements….
If you use a single mic position you could validate some LF phase up to about 300hz , but much above that phase delay will always get in your measurements… a little bit.
Up to about 1k you can get pretty good phase validation with a single mic position and a dual ch analysis… REW won’t cut it .
REW is great for getting good measurements and then making a average and setting phase that way or close to speaker measurements.
With REW you need to do an average of many measurements around your head and the measurements must be coherent (vector averages). The issue with the vector average is that the frequency response has serious problems…
So you need to do a 1st session using db averages, or spl averages to do eq work, then the vector average for phase correct only.
I would check out open sound meter it’s a free dual ch analyzer that’s free…. You’ll need an XLR mic and Interface that has a loop back…. Then you can take some Polar averages and get great phase data and great frequency response data , capture the average around your head (9 second temporal average works nice) make impulse , send to REW , then export as text to rephase for work
An fir does exactly what you ask, if you get in a situation where your fir didn’t do what it was supposed to that means an acoustical issue that will always have that phase signature no matter what you do… so again, validation is tough….. but can be done if you understand the nature of mics and phase and phase delay and making sense of “good enough “
Whatever you capture for left and right ch , and change the ir you sent to rephase should be valid within it self ….. as any anomalies will be fixed for that mic position, or it’s average …. It will just be extremely hard to get that exact same sequence unless you have mic that is fixed in one spot while you go directly to validation stage, but then it’s only valid at one tiny spot. So you must make an average of many spots…. So catch 22 , you’ll have a hard time validation but again , read between the lines. Think logically, and don’t ever assume the fir didn’t do what you asked , instead look at why it isn’t showing what you changed and look at your room or setup
Programs like firD can talk directly to smaart and do a validation live as it makes an iterative fir session for you ….
Best of luck , read this thread for more info it’s all there
Last edited:
Hello
I have no problem with measurements. I know how to do that or to do several of them to get an impression.
I do not measure in the house but outside so there is less distortion due to reflections from walls and more direct input from the loudspeaker source.
However fullrange loudspeaker can be measured pretty close to the drivers in the near field.
A backloaded horn however can be measured 50cm in front of the speaker and you can clearly get only the response from the speaker.
I never measure at listening position as this is much more difficult for getting a good response.
I need advice how to use the rephase software.
Is it possible to use a laptop as sound source and adjust the filters in rephase (phase and frequency response) to taste and apply this filter to the music.
How is that done?
I do not import into rephase a measurement but I want just to export an adjusted filter and later I check with measurment software outside of rephase. This should be pretty simple but how is it done?
The measurement and the iteration of listening and frequency/phase correction until the sound is o.k. is my problem.
Oabeieo
thanks for your answer.I have no problem with measurements. I know how to do that or to do several of them to get an impression.
I do not measure in the house but outside so there is less distortion due to reflections from walls and more direct input from the loudspeaker source.
However fullrange loudspeaker can be measured pretty close to the drivers in the near field.
A backloaded horn however can be measured 50cm in front of the speaker and you can clearly get only the response from the speaker.
I never measure at listening position as this is much more difficult for getting a good response.
I need advice how to use the rephase software.
Is it possible to use a laptop as sound source and adjust the filters in rephase (phase and frequency response) to taste and apply this filter to the music.
How is that done?
I do not import into rephase a measurement but I want just to export an adjusted filter and later I check with measurment software outside of rephase. This should be pretty simple but how is it done?
The measurement and the iteration of listening and frequency/phase correction until the sound is o.k. is my problem.
FIR filters need to be convolved with the input, this does not lend itself to easy real time adjustment, particularly if phase correction is being used as there needs to be delay added to make the filters causal and stable.
By importing the measurement as text into rephase you can tweak the controls in real time and see the effect on the measurement to get it how you want. Export the filter load it into your convolver and listen to it. If you don't like it try something else, export, listen until you are happy or give up
By importing the measurement as text into rephase you can tweak the controls in real time and see the effect on the measurement to get it how you want. Export the filter load it into your convolver and listen to it. If you don't like it try something else, export, listen until you are happy or give up
Yeah listen to fluid , man he’s got this down better then me…
But I like to add I would measure in the room you listen and where you listen if your doing a room correction, and linearizarion on everything…
If you have separate fir for each driver then outside or close mic is preferred as you are doing to get the drivers to sum exact at the 1st wave so to speak.
But sounds like you got a pretty good grip on it, your in for a treat 😇
I built already one way fullrange drivers corrected with IIR filters. It sounds perfect after that.
I want to build now back loaded horns and they have a time delay fault below 200 hertz where the horn dominates the output. Will see if it works as there is some crosstalk at 200 hertz.
Here you see measurements of a 25cm fullrange driver with IIR correction. Phase linearity is naturally given here
https://www.diyaudio.com/community/...w-distortion-with-a-2-way.334757/post-6711184
I want to build now back loaded horns and they have a time delay fault below 200 hertz where the horn dominates the output. Will see if it works as there is some crosstalk at 200 hertz.
Here you see measurements of a 25cm fullrange driver with IIR correction. Phase linearity is naturally given here
https://www.diyaudio.com/community/...w-distortion-with-a-2-way.334757/post-6711184
If you can move your ARTA measurements into REW , then simply export as text to rephase that should work
I do That with my smaart measurements as rephase doesn’t like smaart and freaks out on import as text , REW does a nice job simply reformatting it. So I think you’ll have good luck doing that with your arTa stuff
I do That with my smaart measurements as rephase doesn’t like smaart and freaks out on import as text , REW does a nice job simply reformatting it. So I think you’ll have good luck doing that with your arTa stuff
I will see.
But instead of messing with importing. I can just form some filters export them and listen and measure them. Easy to do it. The revolution is that with fir I can change the time domain and I am curious if it is audible.
I had once Jericho Horn with Fostex Fe208 sigma driver. In a direct A B comparison with a bass reflex design you could hear the loss in dynamics in the 100 to 300 hertz range for the horn.
But instead of messing with importing. I can just form some filters export them and listen and measure them. Easy to do it. The revolution is that with fir I can change the time domain and I am curious if it is audible.
I had once Jericho Horn with Fostex Fe208 sigma driver. In a direct A B comparison with a bass reflex design you could hear the loss in dynamics in the 100 to 300 hertz range for the horn.
Hi guys. i have a two question.
1. When we apply PEQ(from ASR data spinohrama On-axis or Listening Window Correct), Is it good to use apply the FIR filter(LinearPhase)? Or It doesnt matter? IIR, FIR
2. Is it always good to apply this Filter linearization first? (Something like My speaker Spec is 24/db LR crossover 2000hz) with PEQ Fir Filter.
1. When we apply PEQ(from ASR data spinohrama On-axis or Listening Window Correct), Is it good to use apply the FIR filter(LinearPhase)? Or It doesnt matter? IIR, FIR
2. Is it always good to apply this Filter linearization first? (Something like My speaker Spec is 24/db LR crossover 2000hz) with PEQ Fir Filter.
Hi @dingdong3
1. minimum-phase, and also probably proportional Q: https://community.roonlabs.com/t/roon-proportional-q-or-constant-q/177913/3
2. the order does not matter, especially if done within the same FIR
1. minimum-phase, and also probably proportional Q: https://community.roonlabs.com/t/roon-proportional-q-or-constant-q/177913/3
2. the order does not matter, especially if done within the same FIR
@pos Thanks for reply. The part about Filter linearization I mentioned was this.
Some people said, They said that it is better to apply Filter linearization unconditionally.
When listening to 2way and 3way speakers, use the IIR filter to correct the room (~200hz) only.(For Correct Some Huge Peak)
I was wondering if it was also positive to apply only Filter linearization like that without overall phase correction.
Some people said, They said that it is better to apply Filter linearization unconditionally.
When listening to 2way and 3way speakers, use the IIR filter to correct the room (~200hz) only.(For Correct Some Huge Peak)
I was wondering if it was also positive to apply only Filter linearization like that without overall phase correction.
I reread your first post, and would like to amend what I wrote when answering question one above: the part about "proportional Q" only applies to corrections published by Amir when reviewing speakers on ASR, as he is using roon. When building your own corrections based on measurements you can use either constant Q or proportional Q as long as you adjust them yourself to fit your needs.
Regarding your new question, I am not sure I understand the difference you make between "filter linearization" and "overall phase correction".
Regarding your new question, I am not sure I understand the difference you make between "filter linearization" and "overall phase correction".
-> https://www.diyaudio.com/community/...s-a-global-phase-linearization-by-fir.393435/
.... so far the answer seem to be: no. But the debate sen to not be quite finished - if the thread is on topic... it seem to wonder in and out
//
Filters Linearization options in rephase will let you easily correct the known filter types that are included. If you know the allpass response of the filter then you could also undo the phase turn with an inverse allpass. This will correct the effects of the crossover filters and make an IIR filter behave like it's linear phase variant.
You can use these in conjunction with minimum phase room EQ without any issue. Whether linearizing the phase of the crossover filters has any significant audible effect will depend a lot on the crossover frequency, filter slope etc.
There are other sources of phase changes from non minimum phase room behaviour that can be corrected for and can have a much more obvious audible effect. It is also possible to make things much worse with this kind of processing so it is not so straightforward to implement.
Have a question about "phase width".
What defines/limits the width of the valid phase definition in Linear Phase filters ?
When I try different band pass XO's, the width of the phase information appears to be tied to the width of the XO's rolloff gated on each side @ 96dB down irrespective of what slope is selected. e.g. a 24db/octave band-pass XO's phase definition is wider than a 48db/octave, which is wider than a 96dB/octave.
Why 1X at 96dB down and not 1.5 or 2X the width ???
Is this a 64-bit numerical limitation, a software design decision or something else ?
Is there any way to make it wider if it is not a numeric limitation ?
Thanks much.
What defines/limits the width of the valid phase definition in Linear Phase filters ?
When I try different band pass XO's, the width of the phase information appears to be tied to the width of the XO's rolloff gated on each side @ 96dB down irrespective of what slope is selected. e.g. a 24db/octave band-pass XO's phase definition is wider than a 48db/octave, which is wider than a 96dB/octave.
Why 1X at 96dB down and not 1.5 or 2X the width ???
Is this a 64-bit numerical limitation, a software design decision or something else ?
Is there any way to make it wider if it is not a numeric limitation ?
Thanks much.