Not only have we had the ability to do so, doing so only works for one specific measurement axis.
It's also something that the vast majority of us aren't sensitive to and those that are sensitive to it can only hear it with specific recordings, with specific sounds at specific times.
Phase linearisation is always touted as being this holy grail, when in fact it's largely pointless. Certainly it doesn't do any harm but it's not exactly necessary either.
It's also something that the vast majority of us aren't sensitive to and those that are sensitive to it can only hear it with specific recordings, with specific sounds at specific times.
Phase linearisation is always touted as being this holy grail, when in fact it's largely pointless. Certainly it doesn't do any harm but it's not exactly necessary either.
I read the patent, which is about a process for developing their FIR file construction and implementation.
I believe we are going to see alot more such patents, as the science progresses and we learn what can and can't hear/improve, and also learn what system latency's work for what particular applications.
I do think the transition to linear-phase filters in speaker design is inevitable now....in almost mass movement soon....
The effects are already easily audible IMHO. Improved transients, more accurate harmonics..... simply improved clarity... are immediate benefits IME.
I completely agree that using phase correction on a full speaker output is next to useless, the way it just tunes to a spot.
But flattening phase, driver by driver, before summation is proving incredibly worthwhile. Then, using linear phase crossovers for the summation, I'm getting the best sound and measurements, both on-axis and off-axis, I've yet to experience.
Just my 2 cents...
I believe we are going to see alot more such patents, as the science progresses and we learn what can and can't hear/improve, and also learn what system latency's work for what particular applications.
I do think the transition to linear-phase filters in speaker design is inevitable now....in almost mass movement soon....
The effects are already easily audible IMHO. Improved transients, more accurate harmonics..... simply improved clarity... are immediate benefits IME.
I completely agree that using phase correction on a full speaker output is next to useless, the way it just tunes to a spot.
But flattening phase, driver by driver, before summation is proving incredibly worthwhile. Then, using linear phase crossovers for the summation, I'm getting the best sound and measurements, both on-axis and off-axis, I've yet to experience.
Just my 2 cents...
I agree. I never spend more than an hour in front of my system before getting ear fatigue when using textbook LR filters. Even though it's not perfect, the Harsch crossover is big improvement. I can spend hours with minimally processed recordings. Benefits for live electric instruments as well. Bass guitar has never sounded so tight and clear.
The description discusses that using FFT's for convolution is much faster than brute force convolution, duh, that's been known forever (and the answer is exactly the same).
Tests have been done that show the audibility to be minimal.
Someone posted a test a while back on here, one sound sample was filter-less. The other samples had been split in two and filtered via filters of various types and order, then recombined. The unfiltered recording obviously had zero phase wrap, with the others having increasing amounts. The net effect was that you couldn't hear it. The xover chosen was at 1kHz.
The phase shift might be more audible at different frequencies, but we aren't particularly sensitive to it.
Phase shift is something completely different from drivers being properly integrated though. And most loudspeakers suffer from poor design.
This is an example of poor design. Simply applying a textbook LR electrical filter (I'm assuming here) isn't good speaker design. You need LR textbook ACOUSTIC filters for things to work correctly. If using a typical two way then you also need to apply the relevant amount of delay to get the drivers properly integrated with regards to phase. Either that or you need to deviate away from the textbook filters, to asymmetric ones, to get the drivers correctly integrated. Then you need to get the tonal balance right with baffle step compensation and the tweeter level.
Simple cone + dome tweeter speakers may never satisfy and cause listening fatigue, even if correctly designed because you need some sort of directivity control to get them to integrate satisfactorily into the room. Then again there are a number of reasons why a speaker can cause listening fatigue, but mine do not and there's no snake oil or special filters involved.
Wow, expensive system: Ultimate accuracy? Listening to the $80,000 Meyer Sound Bluehorn system | What Hi-Fi? The horns or waveguides look like basic 90 x 40 horns...?
I tried an experiment listening to time aligned and non-time aligned speaker system and could hear the difference. Note this is different than linearizng the phase of an existing system that is not time aligned...
Note that time alignment is not for just one location either...
I tried an experiment listening to time aligned and non-time aligned speaker system and could hear the difference. Note this is different than linearizng the phase of an existing system that is not time aligned...
Note that time alignment is not for just one location either...
I should clarify that I did implement textbook acoustic filters with measurements and all. How do you think I was able to implement a quasi linear phase system with IIR filters and delays
My system consists of prosound coaxial 12's with large format compression drivers with coated diaphragms. I extensively measured, eq'd, and setup multiple crossover slopes over a period of three years with the capability to go back to previous presets.
I honestly didn't want to hear a difference. Maybe some people are more sensitive to phase. I apppreciate the scientific method but I have also observed very well respected people in the audio world take a stance on either side of the issue and implement their craft accordingly. I wouldn't go as far as calling any of them liars.
Perception is an interesting thing. Maybe subtle differences in the way people perceive their environment are responsible for this disagreement.
Last edited:
The tests I've seen have seemed to try to compare linear-phase vs IIR...on speakers that weren't linear-phase to begin with. I hear a big difference between trying to linearize phase on a non-linear speaker, vs a speaker that is linear-phase from the get-go. I think the latter is analogous to a 'correct lens' on its own, the former is using a 'deformed lens' to correct a 'deformed lens'.
Pardon the repetition, but using FIR to correct drivers on an individual basis as a first step in speaker tuning, before even thinking about tying them together, and then using linear-phase crossovers, is what I mean by a linear-phase speaker. I think possibly this is what Meyer and many other are starting doing.... It's a whole different animal ...kinda like mitchba said i think...
No, the design tools for that have existed for a number of years as well. If that is what Meyer and others starting doing......then they're pretty late to the game.
Anyways, the audibility of phase distortion has been discussed ad nauseum on this and many other forums. 'Probably should resume focus on just this Meyer implementation.
Dave.
Last edited:
Sorry to have to simply disagree with most of your points...
Sure the tools, the science, has been around...but I don't think the marketability has (perhaps because of people blindly accepting previous studies
, or the willingness to accept latency when it doesn't matter. Nor has the low cost necessary to implement multi-way FIR existed, until cheaper dsp and cheaper multi-amp capability. And again, IMHO phase distortion audibility has been evaluated via a fundamentally flawed platform.....'wrapped phase' speakers attempted to be taken straight....or simply playing degrees of wrapped vs unwrapped material, on the same (probably wrapped) system.
A better test for phase audibility IMO, would be to start with a linear-phase speaker such as the Meyer, and then introduce phase wrap to it. I like to introduce phase wrap at the crossover level, by changing linear-phase crossovers to IIR's...all types/orders..... and listen. It's why I think linear-phase is desirable audibly.
But to return to the Meyer implementation
Do you see anything new in the patent?Of course !
But linear-phase can only be FIR...yes?!
As an aside, I think one of the biggest benefits of FIR filters, even for those who poo-poo linear phase, is the ability to add nearly unlimited minimum phase filters into the FIR file.
It really looks like they developed a midbass module to augment the Acheron Designer and then made the horn blue to justify giving it a new name. Which is why the marketing reads like, well, marketing.
I cannot speak for other people but I performed the critical comparisons of the different filter types on headphones and couldn't hear anything vs the unfiltered original sample.
Big difference this was not, most likely imperceptible, more likely.
Linear phase is linear phase though. If you've integrated your drivers using standard filters types and their associated phase shifts, then applied a phase linearising FIR filter. Or used phase compensated linear filters to start with. If the end result is a phase linear system, then it doesn't matter how it was arrived at.
Can you give me an example of some loudspeaker design software that can implement, design and output FIR filter coefficients? Even if you were to apply a phase linearised LR4 electrical filter to a driver, you'd still need the filter to take into account the drivers own natural roll off + associated phase shift and compensate for it.
Thanks for the clarification. If you are working with big coaxials then you've got directivity control to start with.
I mean different filters do sound different when you're implementing them on drivers. It's very difficult to end up with a set of drivers that are actually suitable, for trying different filters, so that you're actually comparing the filters and not something else. Usually something will change, along with the filters, that you can attribute a change in sound to. The off axis changing, the distortion profile changing, breakup being less well attenuated, using a driver further away from it's pistonic range of operation, even a minute change in the axial response somewhere, 0.5dB etc. Then you're hearing other things, not the filter.
I've tried this myself, you can read about it here... FSTNT1
Enough to be entirely happy that if filters do impart any difference to the sound that they are so small as to be considered trivial.
Different filters will sound different when crossing over drivers that have limitations and physical properties. You then choose the filters that will allow the drivers to function at their best together. Usually you have very little wiggle room for 'what's best' without sacrificing something or other.
Your coaxials, for example, would necessitate a crossover at about 1200Hz so that you'd get a good directivity match between the cone and the compression driver in the throat of the cone. You could go lower without adversely affecting the off axis response, but the tweeter would probably croak. Going higher is a compromise between tweeter linearity and your off axis response, to the detriment of the speakers overall performance. Ie, if the compression driver can't do 1200Hz, replace it with one that can etc.
What coax and tweeter were you using?
Ah yes. I remember those. Lots of good info on your website.
I'm using Ciare's NDCX12-1.4. I bought them a couple years before the 18 Sound buyout. Not sure if it's changed since or still in production. It gives up top octave performance above 15khz for extended low end performance. The cone is well behaved and extended for its size.
I will admit I haven't tried the headphone tests yet. I will give it a shot and amend my statements if my conclusion changes.
I'm using Ciare's NDCX12-1.4. I bought them a couple years before the 18 Sound buyout. Not sure if it's changed since or still in production. It gives up top octave performance above 15khz for extended low end performance. The cone is well behaved and extended for its size.
I will admit I haven't tried the headphone tests yet. I will give it a shot and amend my statements if my conclusion changes.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.