Thank you FLUID
Okay so I’ll try just that today. We have a snow day in Colorado got to talk the wife into letting me play for a little while.
I stayed up till 3am last night working on the system.
It sounding better than Dirac as far as ambiance, imaging , and harmonic balance. My 2118h midrange are extremely hard to get right.
The differences between left and right responce are so extreme that doing seperate left and right eq de-correlates the phase between left and right.
Where Dirac somehow manages it much better. I really want to figure out how to get that eq work done right and maintain spectral balance in the time domain.
So I’ll be hoping I can get help with that later, for now I’m going continue to get the sub farther nailed down. I’ll start with the sub and work my way up from there.
I’ll get some measurements of the sub just like you described and see if I can make the stopband have better attenuation. I also noticed that and tried adding a iir lowpass and it changed the timing negatively.
Perhaps maybe to do a 24db linearization whilst using a 48db filter or something to that effect.
But I agree, I’ll have to bypass the fir and crossovers and take driver measurements. I’ll see if I can do that today sometime
Here is the link to my drop box that has the .mdat that has sub measurements
close mic and back seat and driver seat and passenger seat. All no processing
the last measurements also did not have processing. I just remembered that I sent my soundcard directly into amps. I do now remember my sub amp (Rockford bd100a1) has a 12db BW dial on it that can not be turned off. I think I have it turned up all the way to like 400hz. Now I know why there's sub timing has been strange. I completely forgot about this.
As far as the sub polar being backwards , I wonder if its the amp polar response or something to do with amp.
I also took some measurements of the 2118H close mic, 1foot away and a driver seat reference mic location for time alignment to left and right. (similar to the 1st Dirac measurement point)
you can see how the 2118h goes to complete s*** at the reference position.
Any ideas welcome
Thanks in advance.
PM me if anyone wants to help me on the side for a consulting fee. I am egar to learn!
If I export the measurement to rephrase with excess phase can I just do a invert time function after my MPEQ (minimum phase eq) work is done?
Dropbox - sub and 2118h.zip - Simplify your life
close mic and back seat and driver seat and passenger seat. All no processing
the last measurements also did not have processing. I just remembered that I sent my soundcard directly into amps. I do now remember my sub amp (Rockford bd100a1) has a 12db BW dial on it that can not be turned off. I think I have it turned up all the way to like 400hz. Now I know why there's sub timing has been strange. I completely forgot about this.
As far as the sub polar being backwards , I wonder if its the amp polar response or something to do with amp.
I also took some measurements of the 2118H close mic, 1foot away and a driver seat reference mic location for time alignment to left and right. (similar to the 1st Dirac measurement point)
you can see how the 2118h goes to complete s*** at the reference position.
Any ideas welcome
Thanks in advance.
PM me if anyone wants to help me on the side for a consulting fee. I am egar to learn!
If I export the measurement to rephrase with excess phase can I just do a invert time function after my MPEQ (minimum phase eq) work is done?
Dropbox - sub and 2118h.zip - Simplify your life
Last edited:
Your close mic measurements show that your drivers behave more as expected, but the phase is still a long way from being minimum phase. See how your phase has a similar shape to minimum up to 70 Hz but is a full turn out of phase, then it goes away. There are no big jumps in the unwrapped phase which usually come from reflections. That could be your mic, your equipment or something, but if you plan on using your measurements for phase correction you should work out what is causing it and if you can remove it.
I have attached an example of a close mic measurement of a single TC9 driver I made. The phase is pretty close to minimum as shown by the grey trace.
When you overlay the close mic and drivers seat sub plots and adjust for level it shows what the cabin is adding. A lot of boost with 80Hz and 27Hz boundary dips. Fairly easy to EQ close to flat with only a few filters.
Your sub seems to extend quite high but your midbass does not go so low. You might want to make your crossover closer to 100Hz and try that.
I have overlaid some textbook curves at 80Hz LR4 and and 1Octave overlap at 100Hz from rephase to show what they look like. If it were me I would EQ the driver to the flat part of the line by removing the peaks above and leave the dips in. If you are using LR24 filters on top of that response you will get quite steep crossovers acoustically. Work out what you are aiming for acoustically overlay that with examples from rephase and see what filters you need to hit those slopes.
I have attached an example of a close mic measurement of a single TC9 driver I made. The phase is pretty close to minimum as shown by the grey trace.
When you overlay the close mic and drivers seat sub plots and adjust for level it shows what the cabin is adding. A lot of boost with 80Hz and 27Hz boundary dips. Fairly easy to EQ close to flat with only a few filters.
Your sub seems to extend quite high but your midbass does not go so low. You might want to make your crossover closer to 100Hz and try that.
I have overlaid some textbook curves at 80Hz LR4 and and 1Octave overlap at 100Hz from rephase to show what they look like. If it were me I would EQ the driver to the flat part of the line by removing the peaks above and leave the dips in. If you are using LR24 filters on top of that response you will get quite steep crossovers acoustically. Work out what you are aiming for acoustically overlay that with examples from rephase and see what filters you need to hit those slopes.
Attachments
Fluid,
Gosh you really know how to make things make sense to me. I have a lot of missing links that your helping so much with. I love how you overlay that. I need to start doing that! Now I see so much better and the smoothing! Aah yes!
I’ve been printing out the responses and looking at them side by side. That is so much easier!
The 2118h is my midrange , I have 2 pair of Stevens audio MB-6 high efficiency midbass drivers playing from 80-200 (sometimes300) I have those two pair of 6.5” in each door (4doors) there about 93db midQ, fs60 ,6mm overhang, awesome little midbass drivers. I have the midbass diialed in. They sound spectacular! The 2118h is a midrange that I am having a hard time getting to sound right that’s why I posted that.
The midbass are Lr4 complimentary 80-200 and the 2118H is LR8 complimentary 200-1.4khz I have noticed much better on and off axis behavior using the LR8on the 2118 and the stage rises a lot with the LR8 on those speakers so I’ve stuck with that.
The horn plays 1.4k LR4 and up, I like the way it sounds with the horn overlapping the midrange a little.
The sub amp I discovered has a built in BW24 crossover that is adjustable from 50-250 and I can’t tell exactly where it’s t it’s a knob with no detents. So I turned the knob up all the way so it should be 250hz BW24. I added to the sub fir a 268hz LR4 linearization and used phase eq to wiggle it into a BW24 linearization. I just realized that the sub had that crossover knob since you said the phase is running backward. So thank you! I forgot that it had that.
The linearization I added I also added the LR4 linearization to the same fir.
The sub sounds much more in time now. I need to remeasure it now and see if the phase is still goofy.
So this is what I need help with the most;
It’s in a car so it’s only for one seat. So there is no worries about other seats or phase issues other than the one seat. That in mind,
What steps do I need to take to correct for the time domain between left and right? Like a room correction, for a gloabal 2ch correction with unequal path lengths?
Let’s say the drivers are all time aligned and the crossovers worked out and everything is sounding good and levels and everything is set. What do I do in the opendrc for a two channel room correction fir assuming everything else is all setup.
What order do I do things and what am I looking for.
Or tell me where I wrong, does it go
1. Take left and right measurements at reference point ( center of head equivalent)
2. Remove any time of flight delay
3. Okay step 3 is where I get stuck , if I eq left and right seperate they de-correlate. How to I do seperate left and right eq without de-correlation. Or do I eq left and right seperate and that gets fixed later with phase correction?
4. So you have let’s say the left channel impulse in REW. You remove time delay
And generate minimum phase , add FDW and export the magnitude and excess phase at txt to rephase
5. Import the impulse into rephase. In rephase I want to see the magnitude and the excess phase with a FDW that was applied in rew.
6.what do I do with the phase!? do I try to make it look like the minimum phase? What is the goal? And how do I do left and right so the transfer functions match each other without screwing up the phase by using minimum phase one one channel and not the other.
I’ve read all the how to’s like Swiss bears how to and a few others but there vague and don’t say much about correlation or coherency between left and right.
Just a basic understanding of the basic steps and I’m sure I will take off and will be able to start to make some great sounding filters. I’ve been stuck at this stage of learning for awhile now and am desperate to figure out what the goal is and some basic how to get there.
I have in REW phase , minimum phase and excess phase. I understand minimum phase enough now, excess phase I understand as the room phase that is added to minimum phase....what about plain phase....is that the sum of minimum and excess I would assume. When doing left and right eq would I simply be able to make the magnitude flat on each respective channel than make the plain phase flat with phase eq for both sides? Would that correct for the diffrent minimum phase eq settings time differences and the room at the same time?
Some basic insight would be helpful, and I wasn’t kidding about a donation.
I called my local sound studio and talked to an engineer here in Denver and asked for tutoring and he said that what I am asking to learn isn’t taught that far in depth and he doesn’t know. It makes me feel like only the ppl that post in this thread and the like are like the only ppl on earth that know how to do this. I can’t find anyone that can teach me.
I also have a bunch of friends on another forum waiting for me to learn this because that also want me to teach them. So there’s a lot of us that are very excited to make the next big step
So much appreciated!
And I’ll check that soundcard. It’s a creative sound blaster usb sound card that has all kinds of outputs and inputs and has a flat response. Maybe I’ll do a soundcard calibration and try again.
Last edited:
I would measure your amplifier at very low volume with REW, less than 0.7V output maybe less to avoid frying your soundcard. Do a search you will find plenty of guides on how to do it.
That will show you what the amp is doing, it should then be possible to create an inverse filter to undo whatever the amp is doing that you don't want.
I have attached a shot of your close mic'ed 2118, this shows good matching to minimum phase before reflections get in the way. That demonstrates that your setup is capable of proper measurements and it is most likely the amp that is causing trouble on the sub measurement.
I have also attached a shot of the drivers seat 2118, with wrapped and unwrapped phase. You can see the jump from the reflections but in the wrapped one you can see the phase still follows the general trend of the minimum phase shape apart from the wraps at reflection points.
I have also done a quick REW auto EQ on that driver to show it isn't that bad with a bit of EQ. Then one with an LR8 @1400Hz.
To set the relative delays start with a single point measurement in your head position. Use REW's acoustic timing reference to find the difference in time of flight from the various drivers. Set those delays in your minidsp. Measure with your crossovers and see what you get.
1 and 2 make sense as above.
3 I don't understand, de-correlation. You want to EQ the left and right drivers so they have as similar a frequency response as you can without trying to fill in the big dips. Make it flat to start with and add the room curve after as global EQ. For now I would only use phase correction to undo the phase turn of your IIR crossovers through filters linearization in rephase.
If you have time aligned the drivers properly and undone the phase turn from your crossovers you should be pretty close to a good step response and minimum phase without anything else.
Then you can measure the left and right channels at the reference point to see what you have got. Apply a frequency dependent window of 15 cycles or less and see what it looks like. Compare the measured phase to the generated minimum phase. Then decide if you think more phase correction is a good idea.
As above make them as similar as you can and as close to flat without filling big dips. Example in the auto EQ screenshot.
The goal is sound that you are happy with
Excess phase is anything other than the minimum phase, it can be from reflections but it could be other sources of non minimum phase behaviour.
What you call plain phase is the measured phase. If your measuring setup is good that should be a representation of what you have. Make sure to include your mic calibration file in your REW measurements. I don't have it so my screenshots don't take it into account.
A speaker is a minimum phase device. If you use minimum phase EQ to correct the amplitude it will correct the phase at the same time.
There are not many situations where trying to correct the phase from reflections will work well. You need a good understanding of what the problem is to decide whether you should try. If not leave it alone.
Any form of phase correction should be the last step. Frequency response, time alignment and crossover optimisation should be your first priorities.
Recording engineer is probably not the best person to talk to about designing a speaker
Huge amount of information here on diyaudio. Do a google site search of diyaudio for the terms you are trying to get your head around.Attachments
I would measure your amplifier at very low volume with REW, less than 0.7V output maybe less to avoid frying your soundcard. Do a search you will find plenty of guides on how to do it.
That will show you what the amp is doing, it should then be possible to create an inverse filter to undo whatever the amp is doing that you don't want.
I have attached a shot of your close mic'ed 2118, this shows good matching to minimum phase before reflections get in the way. That demonstrates that your setup is capable of proper measurements and it is most likely the amp that is causing trouble on the sub measurement.
I have also attached a shot of the drivers seat 2118, with wrapped and unwrapped phase. You can see the jump from the reflections but in the wrapped one you can see the phase still follows the general trend of the minimum phase shape apart from the wraps at reflection points.
I have also done a quick REW auto EQ on that driver to show it isn't that bad with a bit of EQ. Then one with an LR8 @1400Hz.
To set the relative delays start with a single point measurement in your head position. Use REW's acoustic timing reference to find the difference in time of flight from the various drivers. Set those delays in your minidsp. Measure with your crossovers and see what you get.
1 and 2 make sense as above.
3 I don't understand, de-correlation. You want to EQ the left and right drivers so they have as similar a frequency response as you can without trying to fill in the big dips. Make it flat to start with and add the room curve after as global EQ. For now I would only use phase correction to undo the phase turn of your IIR crossovers through filters linearization in rephase.
If you have time aligned the drivers properly and undone the phase turn from your crossovers you should be pretty close to a good step response and minimum phase without anything else.
Then you can measure the left and right channels at the reference point to see what you have got. Apply a frequency dependent window of 15 cycles or less and see what it looks like. Compare the measured phase to the generated minimum phase. Then decide if you think more phase correction is a good idea.
As above make them as similar as you can and as close to flat without filling big dips. Example in the auto EQ screenshot.
The goal is sound that you are happy with
Excess phase is anything other than the minimum phase, it can be from reflections but it could be other sources of non minimum phase behaviour.
What you call plain phase is the measured phase. If your measuring setup is good that should be a representation of what you have. Make sure to include your mic calibration file in your REW measurements. I don't have it so my screenshots don't take it into account.
A speaker is a minimum phase device. If you use minimum phase EQ to correct the amplitude it will correct the phase at the same time.
There are not many situations where trying to correct the phase from reflections will work well. You need a good understanding of what the problem is to decide whether you should try. If not leave it alone.
Any form of phase correction should be the last step. Frequency response, time alignment and crossover optimisation should be your first priorities.
Recording engineer is probably not the best person to talk to about designing a speaker
THATS IT! ah ha! Okay!!!!!
Oh yes! I finally starting to get it now. Oh I wish it wasn’t Black Friday I would be measuring today. This is so exciting. Thank you FLUID!
So here’s where I’ve been screwing it up this whole time,
I have been using moving mic spectral averaging RTA with pink noise to make everything flat. I need to do time averaging. I have always liked the spacial averaging sound way better than using measurement points and averaging them. For this process that must be my problem as I’m doing or it’s adding the room reflections to the average. I need to take those out.
I can’t wait to try this now. Maybe that is what’s causing the de-correlation.
And what I meant by de-correlate is the center image goes away and the phase coherence between left and right falls apart.
In a car there is radical differences between left and right. If let’s say you have a 8db peak at 200 on the left side and a 8db peak at 400 on the right and there’s a 1.2ms time difference and you use time delay to correct for that (which delay as we all know is only a alignment at some frequencies because the acoustic origin is fixed where the acoustic center can be changed with delay) and try to EQ that the phase between drivers gets changed radically.
But I have a hunch now with time averaging I’m only correcting frequencies that are minimum phase and not non minimum phase. Your overlay massively helps me see that!
I’ve been getting screwed up thinking I need to see the overlay in rephase, but all I have to do is move it all to a straight line and that will make it minimum phase again. I did not understand that! Yes it makes sense now.
I am so excited to try this! I’ll do exactly as you said with the impulse peak alignment in rew for each driver. Yes that seems like an excellent approach indeed!
Your a rockstar FLUID , btw there’s some big car audio names filling this thread now. We all are anxiously waiting for this to work right. I’ve been trying so hard to make it make sense. Dirac has a way of doing left and right eq so very nice but it has its issues with tonality that rephase allows me to choose what gets corrected or not or worked on or not. I’m so excited now. I can’t wait to start measurements!
This is it! This is the big one for me. I so much can’t wait to try this.
Last edited:
Maybe see if there is a sub amp that has no filtering going cheap
Using moving mic with an RTA will work better for overall target curve style correction than it will for setting up a crossover. For individual EQ you want to be correcting the speaker and not the environment. There are a number of ways to try and get the room out of the measurement. Spatial averaging or impulse averaging is a good way. As outlined in Swiss Bear's tutorial and discussed before. The other is using a frequency dependant window and that works well enough on a single point measurement.
A strong central image usually comes from having the left and right speakers being very similar in frequency response and sitting in the mid point between them. That is going to be different for a car.
Pure digital delay is the same at all frequencies, the whole point of using time delay in an active system is to correct for the acoustic centre differences of drivers that are not coincident.
Not sure I understand how that helps you understand
Kind of, just don't try and EQ the big dips out and you will be correcting the parts that are behaving as minimum phase. Minimum phase eq will fix amplitude and phase at the same time in that situation.
Start with aligning all the left drivers together, then the right together, at that point you can decide on a relative delay between left and right channels which may help you to get the imaging where you want it to be. A sub can be hard to align in this way because it does not have a sharp impulse, some trial and error might be needed. The step response will show if you have it aligned well or not.
When you have the drivers aligned and EQ'd individually, dirac will have a much easier time, so it might be worth trying it again when you have done some manual tweaking.
That’s fantastic! Yeah I will definitely not forget any of this and I’m sure I’ll be going back and re-reading this along the way along with Swiss bears how to.
So I was under the impression that with unequal path lengths even with signal delay used for time alignment because of the unequal path lengths there would be comb-filtering and because of the physical distance the time alignment only works at frequencies greater than 1 wavelength compared to the path length differences (especially between left and right channels)
For example , speaker pair that plays midbass and/or midrange, the left is let’s say 1/2 wavelength of let’s say 300hz and the right side is one full wavelength of 300hz.
You set the digital delays so both left and right arrive to you at the same time.
The left side starts to play 300hz and arrives at -180 and the right arrives at +180 however they arrive at the same time. The physical distances cause the comb-filtering and the use of the delays is what makes the comb filters emerge.
In this instance (the way I understand it to be and please correct me If I am wrong), the right side if the use of MP-EQ to bring the level down so that frequency is not right side biased and where let’s say 600hz (it’s oactave) would flip where the driver side is at +180 and pssanger at -180. This seems to be a issue and the way I understand things to be and why I see this type of interaction in the mid-bass.
One might think; well no , actually if there’s a 1ms path length difference and you delay 1ms the wavefronts all arrive in phase the same as if you had equal path lengths. (Which is the argument I also understand). Which may be the case, but want happens at 1/2 octaves and octaves above or below the frequencies where there both left and right are at +180 or whatever. It seems to me that the physical offset causes comb filters and trying to eq them to have the same response causes massive phase problems.
And these problems must have eq. The differences in amplitude between left and right are so severe it has to be corrected. So this is where I’m really putting my faith into getting the system to be minimum phase and seeing how that affects the ability to eq the left and right without these issues.
I know everyone says don’t boost the dips. And I wouldn’t want to, however the comb filters are so many from a 4way and each pair suffering from the above issues at all different frequencies because of different pairs being placed in different spots, this almost makes eq work on some of the dips a must.
So in car audio we cut the peaks and don’t boost into anything until we get into the 2db range. If we use the same eq on both channels and just eq the sum of L and R the system sounds great. Once left right eq is applied some frequencies can tolerate it (highs) and some make the speaker go out of phase and the center vanishes.
So again. I’m really counting on this to work. I can’t wait to start measurements. I’ve read all the Dirac pages and they seem to be doing the same stuff rephase is doing. It’s all the same talk. Getting a minimum phase system. I so much want to be able to do my own Dirac correction except do it my way the way I like the eq work done and .....nuff said
I’m actually a car Audio installer by trade and manage a very busy store, so this weekend in retail I’m cooked. I won’t be able to do measurements till at least Monday or Tuesday. And I have a brand new set of beyma 8G40s coming to replace the 2118H (should be here tomorrow) so few more days and it’s on!
Thanks again for the help. I can’t wait to try this
So I was under the impression that with unequal path lengths even with signal delay used for time alignment because of the unequal path lengths there would be comb-filtering and because of the physical distance the time alignment only works at frequencies greater than 1 wavelength compared to the path length differences (especially between left and right channels)
For example , speaker pair that plays midbass and/or midrange, the left is let’s say 1/2 wavelength of let’s say 300hz and the right side is one full wavelength of 300hz.
You set the digital delays so both left and right arrive to you at the same time.
The left side starts to play 300hz and arrives at -180 and the right arrives at +180 however they arrive at the same time. The physical distances cause the comb-filtering and the use of the delays is what makes the comb filters emerge.
In this instance (the way I understand it to be and please correct me If I am wrong), the right side if the use of MP-EQ to bring the level down so that frequency is not right side biased and where let’s say 600hz (it’s oactave) would flip where the driver side is at +180 and pssanger at -180. This seems to be a issue and the way I understand things to be and why I see this type of interaction in the mid-bass.
One might think; well no , actually if there’s a 1ms path length difference and you delay 1ms the wavefronts all arrive in phase the same as if you had equal path lengths. (Which is the argument I also understand). Which may be the case, but want happens at 1/2 octaves and octaves above or below the frequencies where there both left and right are at +180 or whatever. It seems to me that the physical offset causes comb filters and trying to eq them to have the same response causes massive phase problems.
And these problems must have eq. The differences in amplitude between left and right are so severe it has to be corrected. So this is where I’m really putting my faith into getting the system to be minimum phase and seeing how that affects the ability to eq the left and right without these issues.
I know everyone says don’t boost the dips. And I wouldn’t want to, however the comb filters are so many from a 4way and each pair suffering from the above issues at all different frequencies because of different pairs being placed in different spots, this almost makes eq work on some of the dips a must.
So in car audio we cut the peaks and don’t boost into anything until we get into the 2db range. If we use the same eq on both channels and just eq the sum of L and R the system sounds great. Once left right eq is applied some frequencies can tolerate it (highs) and some make the speaker go out of phase and the center vanishes.
So again. I’m really counting on this to work. I can’t wait to start measurements. I’ve read all the Dirac pages and they seem to be doing the same stuff rephase is doing. It’s all the same talk. Getting a minimum phase system. I so much want to be able to do my own Dirac correction except do it my way the way I like the eq work done and .....nuff said
I’m actually a car Audio installer by trade and manage a very busy store, so this weekend in retail I’m cooked. I won’t be able to do measurements till at least Monday or Tuesday. And I have a brand new set of beyma 8G40s coming to replace the 2118H (should be here tomorrow) so few more days and it’s on!
Thanks again for the help. I can’t wait to try this
We aren't in a car audio forum here and I have no first hand experience with car audio, other than driving in different cars and being disappointed by the sound in most
That being said the underlying physics and acoustics involved does not change because you are in a car, but it does change significantly how well any of the strategies that make better home speakers will translate to a car. You'll have to experiment with that yourself to see what you get.
There are two places where you might consider using delays. The first is to try and align the speakers forming the left (or right) channel together. The second is trying to delay the left or right channel to overcome the fact that you are not sitting in the centre between the channels. The first one should be more straightforward, the second one is where what you are describing is more likely to appear.
You can try and fill the edges of the dip by using low Q EQ and that might work to improve things. Only way to know is to measure and see what you get with different amounts.
That being said the underlying physics and acoustics involved does not change because you are in a car, but it does change significantly how well any of the strategies that make better home speakers will translate to a car. You'll have to experiment with that yourself to see what you get.
The delay changes when the speaker will start it does not alter the physical path length and it will not change change any comb filter pattern caused by a driver interacting with it's environment. You could get a change from separate drivers interacting with each other. That is why it is a good idea to set the delays first on each channel and see how that affects the overall response on each channel. If the drivers are not time aligned then applying textbook crossovers and linearizing their phase will not result in a textbook combination. Measure the result after time aligning and applying crossovers. Do you get what you thought you would? If not why not.
I don't understand what you are getting at.
There are two places where you might consider using delays. The first is to try and align the speakers forming the left (or right) channel together. The second is trying to delay the left or right channel to overcome the fact that you are not sitting in the centre between the channels. The first one should be more straightforward, the second one is where what you are describing is more likely to appear.
I don't understand why you would use EQ, if a speaker is more dominant because of position then changing the level would seem to be the solution.
I don't really understand this either.
Why not change the level if it's a level problem
That is why you need to look closely at the measurements and compare close mic'ed to listening position. Is it a dip inherent to the driver, in which case you should be able to fill it. If it is a boundary dip, a null, then filling it will not work and the dip will just come straight back. You can take a measurement full of nulls and dips from boundary interference and EQ out all of those dips so that they disappear on a computer screen. You get a lovely flat line, your job is done. Then you measure it with the correction applied. Hang on where did all my EQ go
You can try and fill the edges of the dip by using low Q EQ and that might work to improve things. Only way to know is to measure and see what you get with different amounts. If using the same drivers on each side then using the same EQ makes sense as you are using speaker EQ rather than environment EQ.
I use DRC FIR because I can change everything about it to get the result I want so I'm with you there.
Good luck
I love it! 
Just wanted to throw out that last argument to see what you think of that logic.
And your very consistent, you definitely know a lot about this stuff. I’m so excited to try this.
So than okay, I will follow exactly that and completely re-train my brain and give this a honest go at it. I’ll make sure to be patient and follow exactly that. That makes solid common sense and is in step with everything I’ve read.
I only asked targeted questions to see if doing this type of correction changes anything in the “norm” of tuning considering unequal path lengths.
And yes levels do make a big difference. However some core frequencies in most cars have two or three big comb-filters that need to be dealt with in one way or another. Just wanted to see what your thoughts were and if that changes how to do the process.
So I’ll do exactly what you said. I’ll tune align all the left speakers to a reference mic position, and make the correction , than do the right channel. Than time align left and right after each correction and listen and measure the sum and go from there.
Thanks again Fluid and thank you for the time to teach me. You know how managing 4 dsps can be all with different delays and such to offset for all the diffrent fir and all the time it takes to get something. Now that I have a clue on how to do this I feel like I have a small degree of confidence thanks to you guys. This is super cool. Best community and smartest ppl here, I feel so much gratitude for all the help.
I’ll definitely keep you guys posted on how it turns out.
Just wanted to throw out that last argument to see what you think of that logic.
And your very consistent, you definitely know a lot about this stuff. I’m so excited to try this.
So than okay, I will follow exactly that and completely re-train my brain and give this a honest go at it. I’ll make sure to be patient and follow exactly that. That makes solid common sense and is in step with everything I’ve read.
I only asked targeted questions to see if doing this type of correction changes anything in the “norm” of tuning considering unequal path lengths.
And yes levels do make a big difference. However some core frequencies in most cars have two or three big comb-filters that need to be dealt with in one way or another. Just wanted to see what your thoughts were and if that changes how to do the process.
So I’ll do exactly what you said. I’ll tune align all the left speakers to a reference mic position, and make the correction , than do the right channel. Than time align left and right after each correction and listen and measure the sum and go from there.
Thanks again Fluid and thank you for the time to teach me. You know how managing 4 dsps can be all with different delays and such to offset for all the diffrent fir and all the time it takes to get something. Now that I have a clue on how to do this I feel like I have a small degree of confidence thanks to you guys. This is super cool. Best community and smartest ppl here, I feel so much gratitude for all the help.
I’ll definitely keep you guys posted on how it turns out.
A small update on the rephase.org web page:
Thanks to member Sixto we now have an english translated version of the most excellent pda0 & Bear tutorial from 2017!
Thanks to member Sixto we now have an english translated version of the most excellent pda0 & Bear tutorial from 2017!
Last edited:
noob question: what is the benefit of FIR instead of IIR in terms of audio quality that we hear? indeed technically phase should be better but i have not read exact post saying "wow i can hear something better"
i have not use FIR yet on my 4 way setup using 2 minidsp 2x4hd. i assume giving more tabs on Sub channel will have more benefit and combine it with Tweeter channel. on another minidsp i can combine Midbass and Midrange
currently i have 3 xml file: 1 xml for 4ways which i can copy to both board, another 2 xml file for Sub-Tweeter and Midbass-Midrange
still trying to decide whether should i jump into FIR hassle or not
i have not use FIR yet on my 4 way setup using 2 minidsp 2x4hd. i assume giving more tabs on Sub channel will have more benefit and combine it with Tweeter channel. on another minidsp i can combine Midbass and Midrange
currently i have 3 xml file: 1 xml for 4ways which i can copy to both board, another 2 xml file for Sub-Tweeter and Midbass-Midrange
still trying to decide whether should i jump into FIR hassle or not
Neither filter type is 'better' or 'worse' in terms of audio quality. The phase is not better with an FIR filter, an FIR filter can be made to have its phase be independent of the magnitude response.
An IIR filter has no latency, an FIR filter has latency based on the filter length (taps) and the position of the impulse within the filter which can be a problem for live sound and lip sync with video.
An FIR filter can introduce varying levels of pre-ringing.
An FIR filter is good for things such as impulse response correction, correcting the phase wrap of crossovers, adjusting phase at the listening position and for very fine grained EQ that is difficult with other filters.
FIR filters are not magic bullets, they can be very useful but you need to know what you are trying to achieve with them as it is just as easy to make things worse.
An IIR filter has no latency, an FIR filter has latency based on the filter length (taps) and the position of the impulse within the filter which can be a problem for live sound and lip sync with video.
An FIR filter can introduce varying levels of pre-ringing.
An FIR filter is good for things such as impulse response correction, correcting the phase wrap of crossovers, adjusting phase at the listening position and for very fine grained EQ that is difficult with other filters.
FIR filters are not magic bullets, they can be very useful but you need to know what you are trying to achieve with them as it is just as easy to make things worse.
@fluid
my assumption that FIR is better seems wrong, i just think that sharc dsp processor capability might be not fully utilized with IIR but you gave me those points
another issue that you mention is important to me about time delay with FIR because i also use it for watching bluray.
i think that i should hold it for a while until i finish building 2nd loudspeaker which will be passive crossover, then i will return to FIR option
my assumption that FIR is better seems wrong, i just think that sharc dsp processor capability might be not fully utilized with IIR but you gave me those points
another issue that you mention is important to me about time delay with FIR because i also use it for watching bluray.
i think that i should hold it for a while until i finish building 2nd loudspeaker which will be passive crossover, then i will return to FIR option
16 Banks
Just wanted to say I like the "16 banks" concept in RePhase for PEQs that I had originally overlooked. They make experimentation quicker and easier. I suppose they can be used in a variety of ways for various reasons, but this is how I have been using them.
I have been using REW's AutoEQ and saving the PEQs off in XML format twice to create 2 duplicate correction files. I then delete the PEQs under Schroeder in one XML file and delete the PEQs above Schroeder in the other XML file. I then load the under Schroeder PEQs in an odd numbered bank and the above Schroeder PEQs into an even numbered bank. From there, I can easily generate minimum-phase, mixed-phase and linear-phase corrections for quick comparisons by just toggling the bank's phase pull down option before generating filters.
If anyone has other uses for the multiple banks, please share.
Just wanted to say I like the "16 banks" concept in RePhase for PEQs that I had originally overlooked. They make experimentation quicker and easier. I suppose they can be used in a variety of ways for various reasons, but this is how I have been using them.
I have been using REW's AutoEQ and saving the PEQs off in XML format twice to create 2 duplicate correction files. I then delete the PEQs under Schroeder in one XML file and delete the PEQs above Schroeder in the other XML file. I then load the under Schroeder PEQs in an odd numbered bank and the above Schroeder PEQs into an even numbered bank. From there, I can easily generate minimum-phase, mixed-phase and linear-phase corrections for quick comparisons by just toggling the bank's phase pull down option before generating filters.
If anyone has other uses for the multiple banks, please share.
functionality of rePhase
Hello guys,
I am a german student and I am currently working on a Multiroom speaker system, which should provide a room correction function. Therefore I need to know a few things about the functionality of rePhase, so it would be great, if some of you could answer me some questions about it.
First of all, I am curious about what happens to the target response, when I increase one particular frequency in rePhase by hand, because when I do so, not just the amplification of the one particular frequency increases, but several surrounding amplifications increases as well. So, first of all i guess this is the case, because rePhase just calculates the expected target response and rePhase expect the amplification of the surrounding frequencies to increase as well. Now, I wonder why this is the case so, why do the surrounding amplifications change?
And second of all, I wonder how rePhase calculates the expected changing of the surrounding amplifications. I would it expect to be a linear interpolation, but I am not sure. Also interesting to know would be how big the area around the increased amplification is, which rePhase also changes and how rePhase determines how big this area is.
Thanks in advance.
Hello guys,
I am a german student and I am currently working on a Multiroom speaker system, which should provide a room correction function. Therefore I need to know a few things about the functionality of rePhase, so it would be great, if some of you could answer me some questions about it.
First of all, I am curious about what happens to the target response, when I increase one particular frequency in rePhase by hand, because when I do so, not just the amplification of the one particular frequency increases, but several surrounding amplifications increases as well. So, first of all i guess this is the case, because rePhase just calculates the expected target response and rePhase expect the amplification of the surrounding frequencies to increase as well. Now, I wonder why this is the case so, why do the surrounding amplifications change?
And second of all, I wonder how rePhase calculates the expected changing of the surrounding amplifications. I would it expect to be a linear interpolation, but I am not sure. Also interesting to know would be how big the area around the increased amplification is, which rePhase also changes and how rePhase determines how big this area is.
Thanks in advance.
Hello gadut, hello fluid,
I think the IIR vs FIR is not only a matter of phase or delay, but more a matter of (CPU/DSP) power and limitations.
As a matter of fact FIR can do *everything* IIR can, including zero delay minimum-phase corrections (with time domain convolution at least), with zero drawback (beside power, but this is relative when you see the power consumption of an openDRC for example).
Yes, you can do FIR correction with zero delay (that is what I am currently running on my speakers).
And of course FIR can also do things IIR cannot do like linear-phase corrections, or very complex (or steep) corrections that would imply a great number of biquads (which brings complications I will cover in the next paragraph), or even purely temporal corrections (reflection cancellation, reverb, etc.).
And this brings another point where FIR is at an advantage compared to IIR: quantization errors.
IIR biquad implementations do generate errors because of the recursive nature of the filter scheme.
Even high precision implementations will generate quantification noises if the Q is high enough and/or the frequency low enough (ie long ringing, implying more accumulated errors). The level of this noise depends on the precision of the calculations, but the more filters you add the more noise you get, and this kind of noise is not as easily masked as harmonic distortion and the like (eg a single LF tone can generate a fullband noise).
In contrast FIR convolution is a direct calculation (even in its frequency domain form), does not depend on the correction itself, and not prone to errors. Errors will also not accumulate when adding multiple filter points. You can use thousands of EQ points in a FIR correction without increasing errors.
One last advantage of FIR vs IIR is correction portability: when importing a correction from one IIR crossover unit (or software) to another you never really know what you will get: filter conventions might vary (constant Q vs proportional Q, fc position for shelving, etc.), as well as their technical implementation (frequency response differences depending on sampling frequency in the UHF, as well as many strange behaviors).
FIR on the other hand is very predictable: once you have designed a correction (using rephase or another FIR tool) you can implement it in any convolution engine and get a predictable result based on the number of taps at hand.
This is an often overlooked point, but major nonetheless as many people are using corrections made for a given processor in another, and get frequency response variations that they often associate with audio qualities or defects of a unit, whereas it is "only" a difference in transfert function.
In short, if you have the processing power FIR only has advantages compared to IIR
Hello emailtim
Interesting use of this feature, thanks for sharing
As for other uses, I know some people have been using multiple banks (all of them even!) for a single correction, trying to refine the correction to the fraction of dB. I cannot say I support this kind of use as over-correction can cause more harm than good (eg correcting measurement artifacts, or position-dependent defects like comb filtering), but if done based on a reliable (set of) measurement(s) and with good analysis then it is probably a valid strategy if one has the time and patience to do so
I tend to only use a limited number of EQ points, but spread them across banks depending on the target of the correction (anechoic response, room adaptation, loudness compensation, driver to driver variation correction, etc.).
One thing that could help in your case (and mine as well) would be to be able to give a different title to each bank, to explicitly explain what it's intend is.
Right now the only way of documenting the use of each bank would be in the global "note" entry in the general tab.