Phase inversion patent

[Cosmik]

Seen here:

...following several years of research, Meyer Sound has now been awarded a United States patent on proprietary digital technologies that enable its Bluehorn studio monitor system to reproduce complex musical signals with absolutely flat frequency and phase response across the full audio bandwidth, an unprecedented achievement in high-power studio monitors.

The new patent, number 9,992,573 B1, is formally titled, "Phase inversion filter for correcting low frequency phase distortion in a loudspeaker system." The award document outlines the digital signal processing techniques involved in canceling out the phase anomalies inherent in all loudspeaker systems due to the physical mass of the loudspeaker drivers and resonance of the loudspeaker enclosures. The patented technology restores the original phase relationships, even at the lowest audio octaves, by applying an inverted phase response.

The patent documentation is here:
https://patents.google.com/patent/US9992573B1/en?oq=9992573

Interesting, because it does almost exactly what I do for my speakers:
Make a measurement of the driver, smooth the phase correction in terms of octaves (e.g. 1/3 octave smoothing), then apply the inverse to the signal in order to pre-correct the phase. Intuitively, smoothing the correction is the obvious thing to do to avoid correcting 'noise' in the measurement.

To me, it is the 'obvious' approach to an automated correction system. I even have the same 'pragmatic' measures like limiting the lower frequency range, windowing the impulse response to generate a practical real world filter, etc. I think that the stuff in the patent is 'obvious' once you set out to correct the phase of a speaker automatically.

I wonder how many existing systems out there do the same thing. If I had been selling my system for all these years would I now be open to being sued retrospectively? Would freely-distributed ('published') code be an automatic defence against being sued? (I doubt it because the code would have to be interpreted by someone in order to verify that was the case). If I had published an article about my system in great detail, would that be an automatic defence? If I had a viable defence, would that invalidate the patent?

I, perhaps, have a subtle difference in my system. They show this image in the patent:

In my system, the crossover is DSP, and the phase correction 'lives' in the same filter as the crossover. The net result would be numerically the same (give or take rounding errors, etc.) as having a separate phase inversion followed by a normal crossover filter but, technically, my system takes advantage of using an integrated DSP crossover and effectively saves a stage of filtering. Does this make my system sufficiently different such that it is not covered by the patent? To me, and any other engineer, it would be a subtle if not negligible difference, but in the world of patents, with attorneys (etc.) having to spread themselves across all kinds of technology, maybe it looks like just as big a difference as any other.

Of course, given a filter at 'run-time', no one can tell how it was created, so this patent really just concerns the process of measurement, etc. and the tool for creating the filter.

My gut feel on this sort of thing is that there was no need for them to patent it. They must know that what they are doing is conceptually already well-established and that it is the practical realisation of the tool that is more important than the general idea or the low level DSP specifics; doing a good user interface would be a far bigger task than the DSP.

All currently academic, because I don't have any plans to do anything with my system except listen to it. But at the back of my mind I had maybe got the notion of doing something with it, and one of the features would have been an automated filter creator.

 

[Wombat]

Seen here:

The patent documentation is here:
https://patents.google.com/patent/US9992573B1/en?oq=9992573

Interesting, because it does almost exactly what I do for my speakers:
Make a measurement of the driver, smooth the phase correction in terms of octaves (e.g. 1/3 octave smoothing), then apply the inverse to the signal in order to pre-correct the phase. Intuitively, smoothing the correction is the obvious thing to do to avoid correcting 'noise' in the measurement.

To me, it is the 'obvious' approach to an automated correction system. I even have the same 'pragmatic' measures like limiting the lower frequency range, windowing the impulse response to generate a practical real world filter, etc. I think that the stuff in the patent is 'obvious' once you set out to correct the phase of a speaker automatically.

I wonder how many existing systems out there do the same thing. If I had been selling my system for all these years would I now be open to being sued retrospectively? Would freely-distributed ('published') code be an automatic defence against being sued? (I doubt it because the code would have to be interpreted by someone in order to verify that was the case). If I had published an article about my system in great detail, would that be an automatic defence? If I had a viable defence, would that invalidate the patent?

I, perhaps, have a subtle difference in my system. They show this image in the patent:
View attachment 14312
In my system, the crossover is DSP, and the phase correction 'lives' in the same filter as the crossover. The net result would be numerically the same (give or take rounding errors, etc.) as having a separate phase inversion followed by a normal crossover filter but, technically, my system takes advantage of using an integrated DSP crossover and effectively saves a stage of filtering. Does this make my system sufficiently different such that it is not covered by the patent? To me, and any other engineer, it would be a subtle if not negligible difference, but in the world of patents, with attorneys (etc.) having to spread themselves across all kinds of technology, maybe it looks like just as big a difference as any other.

Of course, given a filter at 'run-time', no one can tell how it was created, so this patent really just concerns the process of measurement, etc. and the tool for creating the filter.

My gut feel on this sort of thing is that there was no need for them to patent it. They must know that what they are doing is conceptually already well-established and that it is the practical realisation of the tool that is more important than the general idea or the low level DSP specifics; doing a good user interface would be a far bigger task than the DSP.

All currently academic, because I don't have any plans to do anything with my system except listen to it. But at the back of my mind I had maybe got the notion of doing something with it, and one of the features would have been an automated filter creator.

It is not unknown for patents to be issued even though prior art exists. Prior art

Challenging the patent can be time consuming and costly.

 

[svart-hvitt]

Seen here:

The patent documentation is here:
https://patents.google.com/patent/US9992573B1/en?oq=9992573

Interesting, because it does almost exactly what I do for my speakers:
Make a measurement of the driver, smooth the phase correction in terms of octaves (e.g. 1/3 octave smoothing), then apply the inverse to the signal in order to pre-correct the phase. Intuitively, smoothing the correction is the obvious thing to do to avoid correcting 'noise' in the measurement.

To me, it is the 'obvious' approach to an automated correction system. I even have the same 'pragmatic' measures like limiting the lower frequency range, windowing the impulse response to generate a practical real world filter, etc. I think that the stuff in the patent is 'obvious' once you set out to correct the phase of a speaker automatically.

I wonder how many existing systems out there do the same thing. If I had been selling my system for all these years would I now be open to being sued retrospectively? Would freely-distributed ('published') code be an automatic defence against being sued? (I doubt it because the code would have to be interpreted by someone in order to verify that was the case). If I had published an article about my system in great detail, would that be an automatic defence? If I had a viable defence, would that invalidate the patent?

I, perhaps, have a subtle difference in my system. They show this image in the patent:
View attachment 14312
In my system, the crossover is DSP, and the phase correction 'lives' in the same filter as the crossover. The net result would be numerically the same (give or take rounding errors, etc.) as having a separate phase inversion followed by a normal crossover filter but, technically, my system takes advantage of using an integrated DSP crossover and effectively saves a stage of filtering. Does this make my system sufficiently different such that it is not covered by the patent? To me, and any other engineer, it would be a subtle if not negligible difference, but in the world of patents, with attorneys (etc.) having to spread themselves across all kinds of technology, maybe it looks like just as big a difference as any other.

Of course, given a filter at 'run-time', no one can tell how it was created, so this patent really just concerns the process of measurement, etc. and the tool for creating the filter.

My gut feel on this sort of thing is that there was no need for them to patent it. They must know that what they are doing is conceptually already well-established and that it is the practical realisation of the tool that is more important than the general idea or the low level DSP specifics; doing a good user interface would be a far bigger task than the DSP.

All currently academic, because I don't have any plans to do anything with my system except listen to it. But at the back of my mind I had maybe got the notion of doing something with it, and one of the features would have been an automated filter creator.

@Cosmik , this is over my head, sorry...but I wondered - involving DSP - will it still be bit perfect?

Having said that...I think your point on copyrights is food for thought. It’s as if much of «the economy» gets tied up in inflexible arragements (like patents) to the extent that the money-smart people profit instead of the real innovators and creators.

 

[Cosmik]

@Cosmik... involving DSP - will it still be bit perfect?

Something of a philosophical question. You can keep your signal bit perfect as far as you can in the chain - so, for example, you might ensure that your source and processing system are slaved to the DAC's sample rate so there is no sample rate conversion or timing adjustments going on. But once you enter the crossover filtering (and especially if applying corrections based on measurements) you are no longer bit perfect.

The question is: in a conventional non-DSP system does 'bit perfection' mean anything? The answer is: only if you think that passive crossover filters and speaker drivers are perfect. In fact, these components are carrying out grievous corruption of the signal in the analogue domain. The aim of DSP is to perform filtering functions and correction of the drivers in the digital domain and get much closer to perfect overall.

 

[Cosmik]

It’s as if much of «the economy» gets tied up in inflexible arragements (like patents) to the extent that the money-smart people profit instead of the real innovators and creators.

It's why I will only ever be an employee. I simply don't have the mental constitution for commercial, finance and legal-based disputes and must always rely on someone else doing that stuff. The idea of having to pay a lawyer $500 an hour with no guarantee of winning in order to resolve a patent issue on something where I might only make a few thousand dollars' profit anyway..?

I wonder how many Kickstarter campaigns fall foul of this issue.

 

[svart-hvitt]

It's why I will only ever be an employee. I simply don't have the mental constitution for commercial, finance and legal-based disputes and must always rely on someone else doing that stuff. The idea of having to pay a lawyer $500 an hour with no guarantee of winning in order to resolve a patent issue on something where I might only make a few thousand dollars' profit anyway..?

I wonder how many Kickstarter campaigns fall foul of this issue.

Now, this is getting seriously OT, but take a look at this Norwegian producer:

https://www.henryaudio.com/?_ga=2.111950148.1771561276.1532768712-80034276.1531041723

It’s an alternative way of doing R&D, plus production.

Maybe @amirm should measure it.

 

[Wombat]

OT: Do you know that modern law is based on a desire to come up with a way to prevent English nobles from killing each other in duels? A system was established where they could use their wealth, money and connections, to sway their peers in their favour. Nothing has changed much re the masses. Keep litigation running until the inferior(asset-wise) folds.

 

[andreasmaaan]

Good on you for beating them to it @Cosmik

I'm not sure what Meyer is supposed to have done differently to merit the patent. The priority date is 2013, but I'm sure I've been reading about this kind of thing since quite a few years prior to that.

Do you see anything new in the way they're doing it? (claim 1, steps a-f)

On a side note, does a phase-correction filter like this produce pre-ringing? I can't imagine how it couldn't, but I might not be imagining well enough..

 

[Cosmik]

Good on you for beating them to it @Cosmik

I'm not sure what Meyer is supposed to have done differently to merit the patent. The priority date is 2013, but I'm sure I've been reading about this kind of thing since quite a few years prior to that.

Do you see anything new in the way they're doing it? (claim 1, steps a-f)

Yes, I'm pretty sure that this kind of thing is fairly standard, but the sorts of things that engineers dream up every day still seem to end up in patents.

I simply don't know how an existing patent like this applies to basic day-to-day engineering ideas.

If my employer, prior to signing a contract with a customer to develop some bespoke product said to me "Just do a quick search through the patents to check we're not infringing anything..." I would rarely be able to say "We're in the clear.".

If it was my intention to fit an external USB socket to charge the device, I wouldn't *know* that I wasn't infringing this patent, for example:
https://patents.google.com/patent/US7986127

It seems to describe the bog standard use of USB to charge a battery, and is something that any engineer might develop in an afternoon to solve an immediate requirement.

On a side note, does a phase-correction filter like this produce pre-ringing? I can't imagine how it couldn't, but I might not be imagining well enough..

If it's working properly it will introduce pre-ringing. But the ringing for the next driver will be complementary, resulting in zero ringing of any kind (pre- or post-) when they add acoustically - not something you can claim for non-DSP speakers. Of course, the sum won't be perfect in all locations, but we are talking about a relatively shallow slope, and a non-issue.

There is a paper here:
http://www.aes.org/e-lib/browse.cfm?elib=14825

Perceptual Study and Auditory Analysis on Digital Crossover Filters
The extensive research on the perceptual attributes of analog filters used for loudspeaker crossover networks does not necessarily apply to digital filters. In this study finite-impulse response (FIR) and Linkwitz–Riley (LR) digital crossover filters were examined for their perceptual artifacts. Subjective tests with headphones and loudspeakers showed that for LR filters the audibility of phase distortion can be predicted by group delay errors. But FIR filters of high order produce audible artifacts because of time smear created by extensive ringing. LR filters of order 8 or less and FIR filters of about 600 were without problems. These safety limits should be respected.

Most sensible people are going to be using a FIR filter order of something like 4...

 

[andreasmaaan]

If it's working properly it will introduce pre-ringing. But the ringing for the next driver will be complementary, resulting in zero ringing of any kind (pre- or post-) when they add acoustically - not something you can claim for non-DSP speakers. Of course, the sum won't be perfect in all locations, but we are talking about a relatively shallow slope, and a non-issue.

There is a paper here:
http://www.aes.org/e-lib/browse.cfm?elib=14825

Most sensible people are going to be using a FIR filter order of something like 4...

Yes, very interesting. That's one of the papers that also informs my use of 4th order acoustic filters.

And totally agree about patents. Here's a light-hearted look at what is quite a real problem

 

[amirm]

@Cosmik , this is over my head, sorry...but I wondered - involving DSP - will it still be bit perfect?

Once you go through either digital or analog crossover, you no longer have a bit perfect signal.

 

[pos]

If it's working properly it will introduce pre-ringing. But the ringing for the next driver will be complementary, resulting in zero ringing of any kind (pre- or post-) when they add acoustically

Correct, but this is not the case if the phase of the high-pass of the system (eg bass reflex rolloff) is also linearized (as seems to be the case in Meyer Sound's approach), as there is no complementary low pass to cancel ringing.
Same also goes for the low-pass on the other end of the system's bandpass (eg antialiasing filters in DACs)

 

[andreasmaaan]

Correct, but this is not the case if the phase of the high-pass of the system (eg bass reflex rolloff) is also linearized (as seems to be the case in Meyer Sound's approach), as there is no complementary low pass to cancel ringing.
Same also goes for the low-pass on the other end of the system's bandpass (eg antialiasing filters in DACs)

I guess this would also go for the high pass filter created by a closed box, but to a lesser extent?

 

[Wombat]

Doers not talkers get patents. Right or wrong that is how it tends to be.

 

[Cosmik]

Doers not talkers get patents. Right or wrong that is how it tends to be.

I may be sceptical of patents, but I am inventor on quite a few of them
But I'm certainly not the assignee.

 

[SIY]

Back in '85, Warren Vidrine (who was chief scientist at Nicolet and my boss) and I came up with essentially this same idea, but our application was to CD (at that time, a pretty new technology). Yes, there was a late night and beer involved. The company eventually decided that because there was no actual evidence of audibility of the phase shifts associated with the anti-aliasing and anti-imaging filters, it was commercially not interesting and they dropped patenting the idea. Nice to see it return.

Up side: as part of Nicolet's due diligence process, this got me a chance to meet and spend a lot of time in discussions with Dick Greiner. He became a major influence on the way I thought about and approached audio, in a sense rescuing me from a lot of silly ideas I firmly believed.

 

[pos]

I guess this would also go for the high pass filter created by a closed box, but to a lesser extent?

Yes exactly, its phase behavior will be equivalent to a 1st order all-pass (180° total phase shift) whereas a tuned box will be a 2nd order all-pass (360° phase shift).
Then you also have to consider potential additional filters (eg subsonic filters to protect drivers from over-excursion under tuning frequency), as well as any electronic device that is not DC-coupled.

 

[Cosmik]

Correct, but this is not the case if the phase of the high-pass of the system (eg bass reflex rolloff) is also linearized (as seems to be the case in Meyer Sound's approach), as there is no complementary low pass to cancel ringing.

Yes, so the phase linearisation is bringing forward some of the 'natural' post-ringing into pre-ringing. My boxes are closed boxes... I'll not lose sleep over it just yet!

 

[restorer-john]

It is not unknown for patents to be issued even though prior art exists. Prior art

Challenging the patent can be time consuming and costly.

One of the best examples of yore was the whole fiasco with Ivor Tiefenbrun and Hamish Robertson.

from wiki:

The similarities between the LP12 and the Ariston RD11 resulted in a patent case: Ariston vs. Linn. The patent was opposed on various grounds, including that 'what was being claimed as new, was in fact old', and that the idea was 'lacking in inventive step' over what was already known. Further grounds of opposition were that the invention had been 'obtained' from Hamish Robertson, and was his original idea rather than that of Jack Tiefenbrun. Jack Tiefenbrun had formed Castle Precision Engineering (Glasgow) Ltd some 15 years earlier. Hamish Robertson had a company called Thermac in 1967, which became Ariston in 1970, and Ariston Audio in 1973. In 1970 Jack's son Ivor formed a friendship with Hamish. In 1971 Ivor made a prototype turntable with a ball bearing, and then went to Israel. While Ivor was away, Hamish changed the ball bearing to a single point bearing. Robertson's company Thermac then ordered forty of the turntables from Castle Precision Engineering Ltd. In 1971, and now operating as Ariston, Hamish showed the turntable under the model name RD11 at the Harrogate show, and set up a distribution network with C. J. Walker and Company. By the end of 1972 relations between Robertson and the Tiefenbruns had broken down. This allegedly led to a threat to Robertson that a copyright action would be brought against him if he had the RD11 turntable made elsewhere than at Castle Precision Engineering. In February 1973 Linn Products Ltd was formed to sell turntables made by Castle Precision Engineering. Robertson left Ariston, which by now had been taken over by Dunlop Westayr Ltd, and became director Fergus Fons Ltd.