Hi pos,
I have done the measurements and I can confirm everything you said: I was able to correct for the open baffle loss with the paragraphic phase and gain EQs and also if I set the frequencies for filter linearization based on actual measurements and not on the design frequencies the result is very good. Thank you!
In fact rePhase is so easy to use and I made a complete crossover using only convolutions designed with it. Also I could correct for the baffle loss with a linear 1st order filter. Excellent software! Well done!
Based on this experience I have an idea that you may find useful: if you are doing an iterative optimization, maybe you can make possible to determine the minimum number of taps needed for a level of precision. Now I did this by hand, but I think you can ask for the desired level of precision and make iterations to try to lower the number of taps.
It may be also helpful to set the length of the impulse in time units instead of samples, in order to be independent of the sample rate. If you want the same precision at 96Khz as at 48Hz you need to double the number of taps, but in fact the impulse is the same length in the time domain.
Christian
I have done the measurements and I can confirm everything you said: I was able to correct for the open baffle loss with the paragraphic phase and gain EQs and also if I set the frequencies for filter linearization based on actual measurements and not on the design frequencies the result is very good. Thank you!
In fact rePhase is so easy to use and I made a complete crossover using only convolutions designed with it. Also I could correct for the baffle loss with a linear 1st order filter. Excellent software! Well done!
Based on this experience I have an idea that you may find useful: if you are doing an iterative optimization, maybe you can make possible to determine the minimum number of taps needed for a level of precision. Now I did this by hand, but I think you can ask for the desired level of precision and make iterations to try to lower the number of taps.
It may be also helpful to set the length of the impulse in time units instead of samples, in order to be independent of the sample rate. If you want the same precision at 96Khz as at 48Hz you need to double the number of taps, but in fact the impulse is the same length in the time domain.
Christian
I'm listening now to a 2-way system using convolation of rephase impulse files as crossover, phase and EQ. Sounds surprisingly clean and clear. What seems to benefit most are recordings in large spaces, like opera. It does seem to pull farther left and right, tho, dimensioning the center phantom image as compared to a passive crossover.
Each driver (and box) was corrected for phase and EQ, then a linear 4th order Linkwitz-Riley filter applied. I suppose a crossover without the linear phase should be built and compared to this one. Can I really hear the phase differences? I think so.
Each driver (and box) was corrected for phase and EQ, then a linear 4th order Linkwitz-Riley filter applied. I suppose a crossover without the linear phase should be built and compared to this one. Can I really hear the phase differences? I think so.
Pano,
Would you post a wave file of impulse responses of system for raw tweeter, raw woofer, and for corrected woofer, corrected tweeter? Are tweeter and woofer measurements made with microphone at single location?
How big are the boxes? What is driver compliment, spacing and crossover point?
Have you explored this set up with steep crossover using FIR filter?
Generating standard Linkwitz-Riley filters for convolution is easy to do.
I also find recordings in large spaces benefiting greatly from correction. My thoughts are that many of these recordings use microphone arrays that capture hall sound with stereo cues.
Would you post a wave file of impulse responses of system for raw tweeter, raw woofer, and for corrected woofer, corrected tweeter? Are tweeter and woofer measurements made with microphone at single location?
How big are the boxes? What is driver compliment, spacing and crossover point?
Have you explored this set up with steep crossover using FIR filter?
Generating standard Linkwitz-Riley filters for convolution is easy to do.
I also find recordings in large spaces benefiting greatly from correction. My thoughts are that many of these recordings use microphone arrays that capture hall sound with stereo cues.
High B.W. Sure, I think I can post the impulse files. Will have to record the "after" file, don't have it yet. Beware, they won't be pretty. 
The boxes are fairly large, 40"H x 30"W x 24" deep. Horn is the Altec 1005, drivers are 416A woofer and 288 driver. I usually run with a tweeter above 6.5K, but wanted to try a simple 2-way. Crossover is circa 700Hz, 4th order, Linkwitz-Riley. Yes, mic was in 1 position 5' from the speaker face, vertically in between the woofer and horn (the listening axis).
I did try some some steep filters a few years back and did not like them. The drivers I use seem to work better with some blend area, maybe a matter of harmonics. But I could try again, for sure.
The boxes are fairly large, 40"H x 30"W x 24" deep. Horn is the Altec 1005, drivers are 416A woofer and 288 driver. I usually run with a tweeter above 6.5K, but wanted to try a simple 2-way. Crossover is circa 700Hz, 4th order, Linkwitz-Riley. Yes, mic was in 1 position 5' from the speaker face, vertically in between the woofer and horn (the listening axis).
I did try some some steep filters a few years back and did not like them. The drivers I use seem to work better with some blend area, maybe a matter of harmonics. But I could try again, for sure.
Good Question. At the moment I'm using my M-Audio Fast Track Pro as the output, it has 4 channels out. The player is JRiver. I just tell JRiver that I want 4 output channels and the convolver script text file does the routing, crossovers and such. I currently have a file for HP and a file for LP, each 24 bit mono wav. JRiver hits the 4 channels with Left Low, Right Low, Left High, Right High, then it's straight out of the Fast Track to the amps. So JRiver is doing the splitting into 4 channels, I tell the convolver which impulse to use and where to route it. Simple, but took some time to figure out.
JRiver also has crossovers built in, they work. No idea of the phase.
JRiver also has crossovers built in, they work. No idea of the phase.
Thanks! That's all I needed to know and more
This is not so easy to do for those of us chained to our turntables. The signal must be digitized before it can be acted upon. From the MiniDSP stable, it looks like OpenDRC could be used:
The rePhase FIR tool | MiniDSP
But it seems like 2 of them will be needed for a two-way. Any other suggestions for a simpler chain (with the option of digitizing the source)?
Btw, I had almost convinced myself that I can hear the correction applied to a BR box phase response. But then I realized I was listening without enabling the correction. Since then, I've been a little skeptical about hearing changes in phase. I can be convinced though
Pano, on a side note, with my new speakers (A7 + SEOS-24), the sound from some of the records is next to untouchable. Even the transformer-modded CS4398 DAC struggles to be as artifact-free as the SL-1200+DL103+Pearl II combo.
This is not so easy to do for those of us chained to our turntables. The signal must be digitized before it can be acted upon. From the MiniDSP stable, it looks like OpenDRC could be used:
The rePhase FIR tool | MiniDSP
But it seems like 2 of them will be needed for a two-way. Any other suggestions for a simpler chain (with the option of digitizing the source)?
Btw, I had almost convinced myself that I can hear the correction applied to a BR box phase response. But then I realized I was listening without enabling the correction. Since then, I've been a little skeptical about hearing changes in phase. I can be convinced though
Pano, on a side note, with my new speakers (A7 + SEOS-24), the sound from some of the records is next to untouchable. Even the transformer-modded CS4398 DAC struggles to be as artifact-free as the SL-1200+DL103+Pearl II combo.
Last edited:
Cool! Glad the speakers are work well for you.
Yes, there is something to be said for the all analog signal chain. CD can sound very, very good certainly, but there is a joy in the good playback of an LP. As you say, artifact free.
It's a struggle I have, too. How to integrate the turntable into the signal chain. I can run it in via an ADC, but kinda hate to. All this rephasing and active crossover work is fun, but it certainly leads to complex systems.
Yes, there is something to be said for the all analog signal chain. CD can sound very, very good certainly, but there is a joy in the good playback of an LP. As you say, artifact free.
It's a struggle I have, too. How to integrate the turntable into the signal chain. I can run it in via an ADC, but kinda hate to. All this rephasing and active crossover work is fun, but it certainly leads to complex systems.
rePhase measurement import fail format
What format files can be imported to rePhase in measurement import?
I had txt file frequency, magnitude, phase separated with "," decimal separator is ".", but only frequency and magnitude are imported to rePhase, no phase data. Fail is exported from HolmImpulse. Phase export settings in HolmImpulse are Degrees and Unwrap phase. I tested also Unwrap phase not selected, same result, no phase imported to rePhase.
What format files can be imported to rePhase in measurement import?
I had txt file frequency, magnitude, phase separated with "," decimal separator is ".", but only frequency and magnitude are imported to rePhase, no phase data. Fail is exported from HolmImpulse. Phase export settings in HolmImpulse are Degrees and Unwrap phase. I tested also Unwrap phase not selected, same result, no phase imported to rePhase.
Hi Peter,
Not a stupid question at all, and in fact I am really sorry I did not write any documentation for this software
First, the comparison between the DCX and the Dolby Lake was a bit of a joke: you *can* obtain a linear phase filter (and acoustical response) with a DCX and rephase, but you will never get the audio quality (better DACs, better filtering algorithms, etc..) and power (brickwall filters, arbitrary slopes and eq, ...) of a Lake with a DCX.
Still, you will indeed be able to linearize the phase shifts introduced by the IIR filters in the DCX (and drivers/box) with rephase, and get good results (especially if you have crossover points under 1khz, which are more sensitive to phase shifts).
To do so you will only need to dial the filters you are using in your DCX in the "filter linearization" tab in rePhase, and generate an impulse for these (ideally you will want to dial your acoustical filters, not only the electrical ones...).
This is similar to phase arbitrator (which happen to have a user manual
Then in order to use the generated convolution (carrying your phase correction), you will need to find a way to introduce a convolution engine in you audio chain, before the DCX.
If your source is a PC (or Mac) then it should be easy enough: there is a lot a convolution plugin that you can use, depending on your favorite audio player).
If you do not use a PC things will be more difficult: you will have to introduce a piece of hardware to do the convolution.
The minidsp openDRC is currently the only device (that I am aware of) that can do such convolutions.
May I suggest that posts like this be added to the readme at Sourceforge?
I noticed that there aren't a lot of downloads for this software, and some instructions would certainly help
I spent about 60 minutes studying various threads just trying to figure out what it is and what h/w is required to run it.
I've been testing this again tonight. Got good measurements and an impulse that seems correct. Been switching it in and out.
On my system is seems very dependent on the recording whether I hear it or not. Hard to predict. When I do hear it, it's always a similar effect. The linear phase gives me a stronger center phantom image and tends to move voices forward and slightly up. Space seems a bit better defined. If there are drums they seem more dynamic, live.
On some recordings I don't hear it at all. I've tried jazz, pop, rock, dance, classical, opera and lounge. Opera and classical seem to consistently reveal the difference, other genres are hit or miss. I can understand why some people say phase isn't audible, or at least not noticeable. It's not night and day, but sometimes a nice improvement. Further listening is in order.
On my system is seems very dependent on the recording whether I hear it or not. Hard to predict. When I do hear it, it's always a similar effect. The linear phase gives me a stronger center phantom image and tends to move voices forward and slightly up. Space seems a bit better defined. If there are drums they seem more dynamic, live.
On some recordings I don't hear it at all. I've tried jazz, pop, rock, dance, classical, opera and lounge. Opera and classical seem to consistently reveal the difference, other genres are hit or miss. I can understand why some people say phase isn't audible, or at least not noticeable. It's not night and day, but sometimes a nice improvement. Further listening is in order.
May I suggest that posts like this be added to the readme at Sourceforge?
I noticed that there aren't a lot of downloads for this software, and some instructions would certainly help
I spent about 60 minutes studying various threads just trying to figure out what it is and what h/w is required to run it.
Hi Patrick,
rePhase has already been downloaded more than 2700 times in the last 14 months. A result I am quite content with considering the complexity and relative peculiarity of the subject at hand...
The post you are quoting only covers one aspect of rephase (phase linearisation of minimum phase crossover).
You can also use rephase to do the actual filtering (and box/driver phase and amplitude linarisation) using one convolution per channel.
You can use either hardware (openDRC and miniSHARC) or software solutions (plugins, mediaplayers, or even offline conversion) to apply your corrections.
The first post of this thread (linked in the presentation page on sourceforge) is kept up to date and tries to cover all the possible use and link to useful resources.