One on One with Kim Ryrie
Disclaimer: The opinions published here are not representative of the Sydney Audio Club Inc.
At our April Sydney Audio Club Meeting, we had the pleasure of listening to the Serhan Swift mu3 floorstanding speakers being paired with the innovative DEQX Gen-4 LS-200 amplifier. Well-known YouTube audio reviewer John Darko gave the following impression after reviewing the DEQX:
“The audible improvements are more pronounced than switching up cables, server/streamer, DAC or amplifier. Such is the power of low-latency group delay correction through DSP. Australians say Maaate, I say Bloody marvellous!”
Below is the chat one of our members 'JK' had with Kim covering several topics including what DEQX is and how it can help improve one’s audio system. We hope you enjoy this interview!
JK: Kim, for the majority of us who can’t, where did you get the skills, the knowledge and the confidence to co-develop something like the Fairlight CMI and now the DEQX? Was it something you learnt from school, was it from reading, or tinkering with a lot of things growing up?
KR: It’s a good question actually, because both Peter (Vogel) and I, we were at school together and we both had been pulling electronic things apart since we could walk, but on separate trajectories. It wasn’t a formal thing, it was just stuff that we picked up. I would have started that at 6, and Peter a similar age. It’s funny, I think you’re sort of born into it. I don’t know how it works. But you seem to be interested in this stuff.
At our April Sydney Audio Club Meeting, we had the pleasure of listening to the Serhan Swift mu3 floorstanding speakers being paired with the innovative DEQX Gen-4 LS-200 amplifier. Well-known YouTube audio reviewer John Darko gave the following impression after reviewing the DEQX:
“The audible improvements are more pronounced than switching up cables, server/streamer, DAC or amplifier. Such is the power of low-latency group delay correction through DSP. Australians say Maaate, I say Bloody marvellous!”
Below is the chat one of our members 'JK' had with Kim covering several topics including what DEQX is and how it can help improve one’s audio system. We hope you enjoy this interview!
JK: Kim, for the majority of us who can’t, where did you get the skills, the knowledge and the confidence to co-develop something like the Fairlight CMI and now the DEQX? Was it something you learnt from school, was it from reading, or tinkering with a lot of things growing up?
KR: It’s a good question actually, because both Peter (Vogel) and I, we were at school together and we both had been pulling electronic things apart since we could walk, but on separate trajectories. It wasn’t a formal thing, it was just stuff that we picked up. I would have started that at 6, and Peter a similar age. It’s funny, I think you’re sort of born into it. I don’t know how it works. But you seem to be interested in this stuff.
Above: Kim Ryrie and Peter Vogel featured alongside the Fairlight CMI
JK: Did you get into trouble often for doing that?
KR: Yeah a lot. Peter used to stick LDRs (Light Sensitive Resistors) into powerpoints so people would walk into a room and turn the lights on, and the powerpoint would blow up. Yeah that sort of thing, that was a simple trick we did.
JK: So this is the kind of thing we’d need to do, to get to your level?
KR: Yeah, this is Electronics 101
JK: I had a feeling you might have done things like that. Whenever I talk to somebody who has remarkable skills at doing something, often when I ask them “how did you get to this stage?” they always say “I used to pull this thing apart, I used to destroy that”
KR: [chuckles] And it was always a fascination, how things work and eventually you’d just work it out and move on. And then at school, we’d make things for free because we never had any money. We would go to junk shops and buy components for nothing, and so we both built up (intially separately) huge stocks of junk, but we could put things together. So when I suggested to Peter that we do the synthesiser thing, we didn’t even think of the fact that it was supposed to cost money to do that. We were just like “that’ll be alright, we won’t worry about that”. It was just complete stupidity beyond belief. But as it turned out, it sort of worked.
JK: And it’s amazing that you both went to the same school
KR: We did, and we both convinced the teachers that we were working on these really interesting science projects. They let us have access to the science blocks so we’d do all these nefarious things, nothing particularly productive. We created things like new dimmers for the stage lighting in the auditorium, we wanted to do solid state dimmers. To do that, we had to convince the teachers that the old dimmers were dangerous, so we poured some iron filings into the dimmers and of course sparks came flying out as a result. The headmaster then agreed to fund our solid state dimming project, we put in 20 solid-state dimmers and we could control it all remotely with a little control panel that we could take down into the auditorium. There was a master switch on it which controlled the giant relay in the lighting gallery. There we set up a mixture of potassium nitrate and magnesium powder, if you remember the old stage performances of the day where the wizard arrives on the stage and there’s this huge puff of white smoke and explosion and all that, well that’s what potassium nitrate and magnesium was. So we gave the honour of turning on the system to the head teacher and when he switched it on, there was this massive explosion up in the lighting gallery. He turned the master switch on, and a huge flash of white smoke came pouring out of the place. So that’s what we did to amuse ourselves at school.
JK: Now, out of all the tracks that featured the Fairlight CMI in it, which one is your favourite and why?
KR: Oh no, I really couldn’t tell you that
JK: There’s got to be one that is your favourite surely
KR: Well I do like Kate Bush’s stuff, probably because she was some of the earlier stuff and Peter Gabriel’s Sledgehammer, things like that. But I wasn’t so much into the music side of it to be honest. We were so busy trying to stay afloat, and just managed about 100,000 dramas a day, that music was almost the last thing you were worried about.
JK: Did you get into trouble often for doing that?
KR: Yeah a lot. Peter used to stick LDRs (Light Sensitive Resistors) into powerpoints so people would walk into a room and turn the lights on, and the powerpoint would blow up. Yeah that sort of thing, that was a simple trick we did.
JK: So this is the kind of thing we’d need to do, to get to your level?
KR: Yeah, this is Electronics 101
JK: I had a feeling you might have done things like that. Whenever I talk to somebody who has remarkable skills at doing something, often when I ask them “how did you get to this stage?” they always say “I used to pull this thing apart, I used to destroy that”
KR: [chuckles] And it was always a fascination, how things work and eventually you’d just work it out and move on. And then at school, we’d make things for free because we never had any money. We would go to junk shops and buy components for nothing, and so we both built up (intially separately) huge stocks of junk, but we could put things together. So when I suggested to Peter that we do the synthesiser thing, we didn’t even think of the fact that it was supposed to cost money to do that. We were just like “that’ll be alright, we won’t worry about that”. It was just complete stupidity beyond belief. But as it turned out, it sort of worked.
JK: And it’s amazing that you both went to the same school
KR: We did, and we both convinced the teachers that we were working on these really interesting science projects. They let us have access to the science blocks so we’d do all these nefarious things, nothing particularly productive. We created things like new dimmers for the stage lighting in the auditorium, we wanted to do solid state dimmers. To do that, we had to convince the teachers that the old dimmers were dangerous, so we poured some iron filings into the dimmers and of course sparks came flying out as a result. The headmaster then agreed to fund our solid state dimming project, we put in 20 solid-state dimmers and we could control it all remotely with a little control panel that we could take down into the auditorium. There was a master switch on it which controlled the giant relay in the lighting gallery. There we set up a mixture of potassium nitrate and magnesium powder, if you remember the old stage performances of the day where the wizard arrives on the stage and there’s this huge puff of white smoke and explosion and all that, well that’s what potassium nitrate and magnesium was. So we gave the honour of turning on the system to the head teacher and when he switched it on, there was this massive explosion up in the lighting gallery. He turned the master switch on, and a huge flash of white smoke came pouring out of the place. So that’s what we did to amuse ourselves at school.
JK: Now, out of all the tracks that featured the Fairlight CMI in it, which one is your favourite and why?
KR: Oh no, I really couldn’t tell you that
JK: There’s got to be one that is your favourite surely
KR: Well I do like Kate Bush’s stuff, probably because she was some of the earlier stuff and Peter Gabriel’s Sledgehammer, things like that. But I wasn’t so much into the music side of it to be honest. We were so busy trying to stay afloat, and just managed about 100,000 dramas a day, that music was almost the last thing you were worried about.
Above left: Kate Bush | Above Right: Peter Gabriel
JK: For those who were unable to attend the Australian Hi-fi show in Sydney or the Sydney Audio Club meeting in April, can you tell these readers what DEQX is and how it can benefit their system sonically?
KR: Sure. Before I even started Fairlight, I did a bit of loudspeaker design, or I attempted it. I was very bad at it. I got hold of a pair of Altec A7s, they’re the voice of the theatre, horn speakers which we used for our band and we used to blow the horn driver at least every month or two. So someone said why don’t you make that active, separate amplifiers for the horn, another separate amplifier for the bass driver, and then you can have a steeper crossover that protects the horn better and it won’t distort as much. Anyway, I did that and it sounded fantastic. But then I sort of forgot about speaker design for 30 years while doing Fairlight.
And then someone came to me, who used to work for us as a programmer at Fairlight, he worked at CSIRO, and said you should come and listen to this chip we just designed, it does finite impulse response correction and they had a little 6-inch loudspeaker and they measured what it was doing. Then they worked out how to compensate its errors. I listened to that and said that’s really amazing. As I was leaving, they asked would Fairlight be interested? I said well Fairlight doesn’t do loudspeakers but maybe we could think about it. I was doing my sums, it takes nearly half a second of latency to do the processing which of course we couldn’t use. I said “let me know when you fix that”.
Long story short, I decided to fade away from Fairlight, it was becoming a very big business and less fun. I wanted to do something new and so I went to these guys, they were called Lake DSP and they came up with this long convolution filtering. They said why don’t you do this speaker, we’re working on the headphone technology (which became Dolby Headphones) and I said ok because I like loudspeakers and making them sound better.
So DEQX stands for Digital Equalization and the X is for crossover. Ideally you have an active system where you’ve got a separate amplifier for each driver, one for the bass driver, one for the tweeter, one for the midrange etc., but you don't have to do that. You can do what we did like for the car speaker, you can just get a traditional loudspeaker and measure it pseudoanechoically. In other words, you don’t want to get the room too involved in the measurement. We really want to hear the detail coming from the speaker itself, because once the reflections in the room corrupt the measurements, it loses the phase information coming from the speaker. That is why, we measure the speaker quite close up, only about a foot or 2 away from the midrange & tweeter driver. That lets us know exactly what’s happening to the timing coherence of all the frequencies throughout the midrange, the high frequency and the upper bass. We don’t worry too much about the lower bass because that’s really room issues and we deal with that when we do a room measurement.
JK: For those who were unable to attend the Australian Hi-fi show in Sydney or the Sydney Audio Club meeting in April, can you tell these readers what DEQX is and how it can benefit their system sonically?
KR: Sure. Before I even started Fairlight, I did a bit of loudspeaker design, or I attempted it. I was very bad at it. I got hold of a pair of Altec A7s, they’re the voice of the theatre, horn speakers which we used for our band and we used to blow the horn driver at least every month or two. So someone said why don’t you make that active, separate amplifiers for the horn, another separate amplifier for the bass driver, and then you can have a steeper crossover that protects the horn better and it won’t distort as much. Anyway, I did that and it sounded fantastic. But then I sort of forgot about speaker design for 30 years while doing Fairlight.
And then someone came to me, who used to work for us as a programmer at Fairlight, he worked at CSIRO, and said you should come and listen to this chip we just designed, it does finite impulse response correction and they had a little 6-inch loudspeaker and they measured what it was doing. Then they worked out how to compensate its errors. I listened to that and said that’s really amazing. As I was leaving, they asked would Fairlight be interested? I said well Fairlight doesn’t do loudspeakers but maybe we could think about it. I was doing my sums, it takes nearly half a second of latency to do the processing which of course we couldn’t use. I said “let me know when you fix that”.
Long story short, I decided to fade away from Fairlight, it was becoming a very big business and less fun. I wanted to do something new and so I went to these guys, they were called Lake DSP and they came up with this long convolution filtering. They said why don’t you do this speaker, we’re working on the headphone technology (which became Dolby Headphones) and I said ok because I like loudspeakers and making them sound better.
So DEQX stands for Digital Equalization and the X is for crossover. Ideally you have an active system where you’ve got a separate amplifier for each driver, one for the bass driver, one for the tweeter, one for the midrange etc., but you don't have to do that. You can do what we did like for the car speaker, you can just get a traditional loudspeaker and measure it pseudoanechoically. In other words, you don’t want to get the room too involved in the measurement. We really want to hear the detail coming from the speaker itself, because once the reflections in the room corrupt the measurements, it loses the phase information coming from the speaker. That is why, we measure the speaker quite close up, only about a foot or 2 away from the midrange & tweeter driver. That lets us know exactly what’s happening to the timing coherence of all the frequencies throughout the midrange, the high frequency and the upper bass. We don’t worry too much about the lower bass because that’s really room issues and we deal with that when we do a room measurement.
Above: DEQX Gen-4 Speaker Calibration and Room Correction Processor
So the first thing DEQX does is it focuses on restoring the timing accuracy through the midrange, which is never perfectly accurate from a loudspeaker because of the electromechanics of the loudspeaker and the passive crossover electronics. They all introduce phase errors. Phase errors means a loss of the original live timing coherence, and only the very best speakers manage to maintain that. We’re so used to hearing loudspeakers sound like loudspeakers, you can always tell because you can hear this smeared timing. We’re not appalled by it because we’re so used to it, sometimes we don’t even notice it but when it goes away, things just sound more natural so that’s the core of what we do.
Now with really good speakers like the Serhan Swift mμ3F speakers that we demoed at the SAC Meeting, they are unusually accurate for a passive speaker and in a way it was stupid of us to use their speakers to demonstrate the DEQX because they’re so good to begin with, but you could have told when we put the DEQX in. We only did a fairly quick measurement of the speakers on the balcony before the show started, you can tell the slight tightening up of everything, the midrange was cleaner, the imaging gets more solid. In their case, it was not dramatically improved, but it was certainly noticeable and as Morris said, the quality that we’ve gotten now in the new generation for DAC hardware which we spent 4 years developing is so good anyway, that it wouldn’t matter if you didn’t even use the DEQX, it’s up there with the top DACs and preamps and streamers. So the DEQX correction and the room correction and the ability to do parametric EQ is almost a freebie.
So the first thing DEQX does is it focuses on restoring the timing accuracy through the midrange, which is never perfectly accurate from a loudspeaker because of the electromechanics of the loudspeaker and the passive crossover electronics. They all introduce phase errors. Phase errors means a loss of the original live timing coherence, and only the very best speakers manage to maintain that. We’re so used to hearing loudspeakers sound like loudspeakers, you can always tell because you can hear this smeared timing. We’re not appalled by it because we’re so used to it, sometimes we don’t even notice it but when it goes away, things just sound more natural so that’s the core of what we do.
Now with really good speakers like the Serhan Swift mμ3F speakers that we demoed at the SAC Meeting, they are unusually accurate for a passive speaker and in a way it was stupid of us to use their speakers to demonstrate the DEQX because they’re so good to begin with, but you could have told when we put the DEQX in. We only did a fairly quick measurement of the speakers on the balcony before the show started, you can tell the slight tightening up of everything, the midrange was cleaner, the imaging gets more solid. In their case, it was not dramatically improved, but it was certainly noticeable and as Morris said, the quality that we’ve gotten now in the new generation for DAC hardware which we spent 4 years developing is so good anyway, that it wouldn’t matter if you didn’t even use the DEQX, it’s up there with the top DACs and preamps and streamers. So the DEQX correction and the room correction and the ability to do parametric EQ is almost a freebie.
Above: The Serhan Swift mμ3F Floorstanding Speaker
JK: That time smearing, is that one of the factors that differentiates real live music from reproduced music via loudspeakers?
KR: Absolutely it is, I mean what else is there? It’s all about phase and timing. Funnily enough when DEQX first started many years ago, everyone was so into frequency response. They never heard of timing response, like group delay errors, so we only talked about frequency response. Oh look at this, now here’s a speaker and we can make it have a flat frequency response. And a flat frequency response done properly is how a speaker should sound, because all that means is it’s just reproducing every frequency at exactly the level it’s supposed to. When you put it into a room, you don’t want flat. When you put it into a room, the room has a different effect which means that you actually do not want a flat frequency response, you want about 6dB more level in the bass end of the spectrum and less at the higher end. With the DEQX, you’ve got parametric EQ anyway so you can do your own tweaking, do multiple bands of parametric EQ to make things sound exactly how you like. The beauty of the system is, for people who are wondering, it just gives you control over what you ultimately hear. Traditionally you buy certain speakers because you like the sound of those speakers, and you get what you’re given. You might have a bass and a treble control and that’s about it. When we do speaker correction, it’s effectively about 4000 groups of frequencies that we control individually, so imagine a 4000-band parametric equalizer that not only corrects the volume, but it also corrects the timing. When I say timing coherence, what loudspeakers tend to do or what all loudspeakers do, you’ll get some frequency groups that are slightly delayed compared to others. And it’s only slight, it can be only 0.5ms for example. That doesn’t sound like much, but when you consider how our ears can detect the difference in timing to a fraction of that, to detect where direction is coming from etc, we inherently notice the slightest issues with timing coherence. Unexpected coherence timing means it’s not sounding live, sounding a bit weird so what we do is we first mathematically split all the audio up into thousands of bands, the bands that are on time are the ones that are only delayed to the degree needed to restore the original phase-linear timing coherence. The ones that are running late are then only delayed to the degree needed to catch up. That’s how it ends up being restored coherently.
JK: So DEQX is essentially fixing the sound on two parameters, one would be the amplitude of the frequencies and the other is the timing of those frequencies
KR: That’s right, exactly. And those are called group delay errors. So there’s frequency response errors and group delay errors, and together they represent the impulse response. Everything in the entire audio chain has some degree of impulse response error introduced, even a speaker cable introduces some degree of impulse response error which the DEQX will compensate for as well. So everything including the amplifier, the cables, all the way through to the transducers, everything gets compensated in that process. It’s just that the speaker is an order of magnitude worse than everything else, because of their electromechanics.
Some audiophiles have said “oh yeah I’ve heard these flat speakers and they measure really flat but they sound horrible and can sound lifeless”. What happens with traditional equalization, is they’re actually introducing phase errors, they’re making them worse so the cure is sometimes worse than the disease. That’s one of the biggest problems we’ve had from day one, with people not understanding. They hear the word ‘equalization’ and they head for the hills, so that’s something that’s been misunderstood.
JK: How does DEQX differ from other products on the market that offer DIRAC?
KR: If your speakers are good already and your main issue is room issues and resonances, and you’ll always have resonances in the room around at about 50, 60 to 80 Hz which are based on the dimensions and height of the room, then Dirac does a pretty good job of cleaning that up through the lower midrange. It’ll do some sort of timing correction, it sort of has a stab at it so I don’t want to suggest that we’re the only ones on Earth that does it, but we do it differently by doing the measurements of the speaker drivers separately. This way, we’re able to get much more resolution, we’re effectively correcting the native speaker’s performance. It’s a double-process with the DEQX, it’s only a single-process with the Dirac.
JK: There are currently 3 products in the DEQX lineup, the Pre-4, Pre-8 and the LS-200. Can you briefly tell us the difference between all 3 and what application each is best suited to?
KR: There’s 2 categories of buyers, our traditional market are the serious users who want to do active loudspeakers. Most of the earlier DEQX units went to people who own active speakers.
If you have a full range speaker and you don’t want to make it active, you’d go with the Pre-4. That has 4 outputs, meaning it has 2 outputs and 2 subwoofer outputs. So the Pre-4 would be for the more traditional sort of speaker person who has their own favourite amplifier and owns a passive loudspeaker.
JK: That time smearing, is that one of the factors that differentiates real live music from reproduced music via loudspeakers?
KR: Absolutely it is, I mean what else is there? It’s all about phase and timing. Funnily enough when DEQX first started many years ago, everyone was so into frequency response. They never heard of timing response, like group delay errors, so we only talked about frequency response. Oh look at this, now here’s a speaker and we can make it have a flat frequency response. And a flat frequency response done properly is how a speaker should sound, because all that means is it’s just reproducing every frequency at exactly the level it’s supposed to. When you put it into a room, you don’t want flat. When you put it into a room, the room has a different effect which means that you actually do not want a flat frequency response, you want about 6dB more level in the bass end of the spectrum and less at the higher end. With the DEQX, you’ve got parametric EQ anyway so you can do your own tweaking, do multiple bands of parametric EQ to make things sound exactly how you like. The beauty of the system is, for people who are wondering, it just gives you control over what you ultimately hear. Traditionally you buy certain speakers because you like the sound of those speakers, and you get what you’re given. You might have a bass and a treble control and that’s about it. When we do speaker correction, it’s effectively about 4000 groups of frequencies that we control individually, so imagine a 4000-band parametric equalizer that not only corrects the volume, but it also corrects the timing. When I say timing coherence, what loudspeakers tend to do or what all loudspeakers do, you’ll get some frequency groups that are slightly delayed compared to others. And it’s only slight, it can be only 0.5ms for example. That doesn’t sound like much, but when you consider how our ears can detect the difference in timing to a fraction of that, to detect where direction is coming from etc, we inherently notice the slightest issues with timing coherence. Unexpected coherence timing means it’s not sounding live, sounding a bit weird so what we do is we first mathematically split all the audio up into thousands of bands, the bands that are on time are the ones that are only delayed to the degree needed to restore the original phase-linear timing coherence. The ones that are running late are then only delayed to the degree needed to catch up. That’s how it ends up being restored coherently.
JK: So DEQX is essentially fixing the sound on two parameters, one would be the amplitude of the frequencies and the other is the timing of those frequencies
KR: That’s right, exactly. And those are called group delay errors. So there’s frequency response errors and group delay errors, and together they represent the impulse response. Everything in the entire audio chain has some degree of impulse response error introduced, even a speaker cable introduces some degree of impulse response error which the DEQX will compensate for as well. So everything including the amplifier, the cables, all the way through to the transducers, everything gets compensated in that process. It’s just that the speaker is an order of magnitude worse than everything else, because of their electromechanics.
Some audiophiles have said “oh yeah I’ve heard these flat speakers and they measure really flat but they sound horrible and can sound lifeless”. What happens with traditional equalization, is they’re actually introducing phase errors, they’re making them worse so the cure is sometimes worse than the disease. That’s one of the biggest problems we’ve had from day one, with people not understanding. They hear the word ‘equalization’ and they head for the hills, so that’s something that’s been misunderstood.
JK: How does DEQX differ from other products on the market that offer DIRAC?
KR: If your speakers are good already and your main issue is room issues and resonances, and you’ll always have resonances in the room around at about 50, 60 to 80 Hz which are based on the dimensions and height of the room, then Dirac does a pretty good job of cleaning that up through the lower midrange. It’ll do some sort of timing correction, it sort of has a stab at it so I don’t want to suggest that we’re the only ones on Earth that does it, but we do it differently by doing the measurements of the speaker drivers separately. This way, we’re able to get much more resolution, we’re effectively correcting the native speaker’s performance. It’s a double-process with the DEQX, it’s only a single-process with the Dirac.
JK: There are currently 3 products in the DEQX lineup, the Pre-4, Pre-8 and the LS-200. Can you briefly tell us the difference between all 3 and what application each is best suited to?
KR: There’s 2 categories of buyers, our traditional market are the serious users who want to do active loudspeakers. Most of the earlier DEQX units went to people who own active speakers.
If you have a full range speaker and you don’t want to make it active, you’d go with the Pre-4. That has 4 outputs, meaning it has 2 outputs and 2 subwoofer outputs. So the Pre-4 would be for the more traditional sort of speaker person who has their own favourite amplifier and owns a passive loudspeaker.
Above: The DEQX Pre-4 is suitable for those with passive speakers and want to use their own power amplifier
The LS-200 does the exact same thing, except it does its own amplification internally, it’s more like an integrated lifestyle.
The LS-200 does the exact same thing, except it does its own amplification internally, it’s more like an integrated lifestyle.
Above: The LS-200 Integrated Amplifier is similar to the Pre-4 but does not require a separate power amplifier
The Pre-8 is for our most serious users, or people who want to upgrade their passive loudspeakers to active, meaning they’ll unscrew their speaker boxes and bypass the passive crossovers inside them and wire up separate amplifiers to the tweeter, to the midrange and to the bass and to the subwoofer. The advantage of this is you’re running short speaker cables to the speaker box. It’s better to run long balanced line-level cables than lots of speaker cables.
The Pre-8 is for our most serious users, or people who want to upgrade their passive loudspeakers to active, meaning they’ll unscrew their speaker boxes and bypass the passive crossovers inside them and wire up separate amplifiers to the tweeter, to the midrange and to the bass and to the subwoofer. The advantage of this is you’re running short speaker cables to the speaker box. It’s better to run long balanced line-level cables than lots of speaker cables.
Above: The DEQX Gen-4 Pre-8 designed for those who want to run active speakers
The other advantage is that there’s less stress on the speaker cables as each cable is not doing as much. The reason you can tell differences between speaker cables, is that you’ve got amps running 40Hz combined with this incredible air happening at 10kHz and you expect it all to sort of hang together nicely. It’s much easier to have one cable that’s handling the top couple of octaves just by itself with no interference, and then another cable that’s only running midrange frequencies. Then you have another cable running 3 octaves of bass, then everyone’s happy as pigs in a proverbial {bleep} and it’s far less stress on the amplifiers, and that’s why everything sounds so much more open and relaxed.
The only reason audio hasn’t been active from day one, is that it’s just a nightmare for marketing, how would you do it? You basically want the retailer to be able to say, well this is an amplifier, it will work with any of these speakers, end of discussion. You need 2 wires, to go from here to there, that’s it. But that has always been a dire compromise if you want it to achieve true high-definition audio and that’s why pro-audio has always been active for as long as I can remember.
JK: When you buy an active speaker, they’ll usually come with an amplifier chosen by the designer as a package. With the Pre-8, you can choose your own amplifier.
KR: Exactly, it’s a done deal. It’s why we are aimed at people that like experimenting. The other thing we haven’t spoken about is that DEQX units comes with microphones that are very accurate. We get our microphones made from a company called Earthworks and they have a microphone now that we tend to supply with our units for the M23R and that is accurate without any mic calibration file to +/- 0.3 dB from 20Hz to 20kHz. In other words, it’s dead flat. But what’s great about that is the impulse response is dead flat, whereas if I had to introduce a calibration file to fix up a cheap microphone, I’m not correcting the phase errors.
The other point about DEQX is that the whole set up is integrated with what we call the DEQX cloud, that’s where all of our computation happens. It leads you through the steps, how to measure each driver individually. Those measurements are then sent up to the cloud and you set where you want the crossover to be. You can see the measurements, how they are in their natural state; that tweeter is rolling off at 1kHz but I think I’ll cross it over at about 2kHz for example, to give it a bit of headroom. Then you do the same with the midrange and you can set the crossover for that, and similarly from the midrange to the bass, and the bass driver to the subwoofer. So all that happens automatically with our measurement system, it’s sent to the cloud, and then it creates this whole filter that gets sent back to the system. Then you have a listen to it, and you say, oh that sounds pretty good. Then you can say but I wonder what happens if I move this crossover down, or up, so of course people love getting into all this crazy stuff and it’s fun because it’s so quick and easy to try different things.
JK: So you can adjust the crossovers using DEQX and bypass the default crossover set by the company.
KR: You can, the (brand of speakers) will usually tell you what their crossover is, and you can completely ignore it if you want. Their crossover would have been set due to limitations that they’ve got with their passive crossover design. For example, they will have a passive crossover typically relatively shallow because you don’t want to do steep crossovers using passive components. It’s extremely expensive and it’s fraught with danger in terms of introducing phase distortions, which means the crossovers that’s tuned to happen in our traditional passive speakers, sort of 2-pole, (12dB per octave) or 3-pole (18dB per octave) so that’s relatively shallow compared to what we will often do. They’re called minimum phase filters which means by definition there are already phase errors introduced.
The only way you get no phase errors introduced is by using linear phase crossover filters, and the only way you can do that in a passive design is with a single-pole filter. Using a single capacitor to the tweeter for example and very simplistic electronics, you have almost no protection for the tweeter from bass frequencies, it will just instantly drive it into crazy territory. You need a steeper filter to protect the tweeter from bass distortions, so what we like to do is to suggest to people use pretty steep linear-phase filters which we can do because we are in the digital domain. We are using convolution to do the filters, which means we can maintain linear phase even with very steep filters like 60dB per octave for example, which is a 10-pole filter. And that’s like a brickwall protection to a tweeter, it will shut the midrange driver crossing over to the tweeter. So with a steep crossover, that midrange driver is going to sound more natural because it won’t start beaming. The tweeter will be completely omni-directional, and the midrange will probably still be largely omni-directional, it will give you a much more natural dispersion of frequencies at that crossover region and it just sounds more natural basically.
JK: In your experience so far, you’ve probably measured a couple of different speakers already. Which speakers have you found to have the lowest distortion?
KR: (Chuckles) Look that’s a good question and it’s very sad by the way, that speaker manufacturers do not publish their distortion. They know how bad it is, when people are used to reading distortions of 0.001% with amplifiers, and their speakers are 1% distortion.
Almost all affordable consumer speakers are about 1% distortion. Now I know all about 1% distortion because the Fairlight had that because it only had 8-bits of audio and that was the best it could do. When you have 1% of distortion, you’re getting about 40dB of resolution. In other words, once I get down to -40dB, I start hearing introduced harmonics that shouldn’t be there. In the case of a speaker, fortunately they tend to be the second order harmonics which isn’t too agonizing to listen to, but it does mask resolution of lower detail. Not far behind, is the third harmonic which is really irritating, and you do start noticing it. So what’s great are speakers that can manage about 0.1% distortion and that means they have 60dB of clean resolution before introduced harmonics start arriving.
An example of that would be the magnesium midrange cones and I’m really focusing on midrange because that’s where it’s critical to try to get the distortion as low as you can. Athough tweeters manage to get to 0.1% and that’s why I suspect a lot of people are into these freakishly high sample rates, because the only way they can hear any difference is if the tweeter has enough high resolution to even make that noticeable and tweeters do tend to have quite good resolution above 2kHz. but getting back to the midrange, the SEAS magnesium drivers for example I think they are the Excel range in the 5-inch, they were a popular one that we used to use in a speaker, I quite like the new Purifi drivers that look as ugly as hell, but all of them without exception can manage 0.1% distortion and it’s just astonishing how they do it, I don’t know how they do it. Another one that does 0.1% distortion are the ceramic drivers that cost an absolute arm and a leg.
The other advantage is that there’s less stress on the speaker cables as each cable is not doing as much. The reason you can tell differences between speaker cables, is that you’ve got amps running 40Hz combined with this incredible air happening at 10kHz and you expect it all to sort of hang together nicely. It’s much easier to have one cable that’s handling the top couple of octaves just by itself with no interference, and then another cable that’s only running midrange frequencies. Then you have another cable running 3 octaves of bass, then everyone’s happy as pigs in a proverbial {bleep} and it’s far less stress on the amplifiers, and that’s why everything sounds so much more open and relaxed.
The only reason audio hasn’t been active from day one, is that it’s just a nightmare for marketing, how would you do it? You basically want the retailer to be able to say, well this is an amplifier, it will work with any of these speakers, end of discussion. You need 2 wires, to go from here to there, that’s it. But that has always been a dire compromise if you want it to achieve true high-definition audio and that’s why pro-audio has always been active for as long as I can remember.
JK: When you buy an active speaker, they’ll usually come with an amplifier chosen by the designer as a package. With the Pre-8, you can choose your own amplifier.
KR: Exactly, it’s a done deal. It’s why we are aimed at people that like experimenting. The other thing we haven’t spoken about is that DEQX units comes with microphones that are very accurate. We get our microphones made from a company called Earthworks and they have a microphone now that we tend to supply with our units for the M23R and that is accurate without any mic calibration file to +/- 0.3 dB from 20Hz to 20kHz. In other words, it’s dead flat. But what’s great about that is the impulse response is dead flat, whereas if I had to introduce a calibration file to fix up a cheap microphone, I’m not correcting the phase errors.
The other point about DEQX is that the whole set up is integrated with what we call the DEQX cloud, that’s where all of our computation happens. It leads you through the steps, how to measure each driver individually. Those measurements are then sent up to the cloud and you set where you want the crossover to be. You can see the measurements, how they are in their natural state; that tweeter is rolling off at 1kHz but I think I’ll cross it over at about 2kHz for example, to give it a bit of headroom. Then you do the same with the midrange and you can set the crossover for that, and similarly from the midrange to the bass, and the bass driver to the subwoofer. So all that happens automatically with our measurement system, it’s sent to the cloud, and then it creates this whole filter that gets sent back to the system. Then you have a listen to it, and you say, oh that sounds pretty good. Then you can say but I wonder what happens if I move this crossover down, or up, so of course people love getting into all this crazy stuff and it’s fun because it’s so quick and easy to try different things.
JK: So you can adjust the crossovers using DEQX and bypass the default crossover set by the company.
KR: You can, the (brand of speakers) will usually tell you what their crossover is, and you can completely ignore it if you want. Their crossover would have been set due to limitations that they’ve got with their passive crossover design. For example, they will have a passive crossover typically relatively shallow because you don’t want to do steep crossovers using passive components. It’s extremely expensive and it’s fraught with danger in terms of introducing phase distortions, which means the crossovers that’s tuned to happen in our traditional passive speakers, sort of 2-pole, (12dB per octave) or 3-pole (18dB per octave) so that’s relatively shallow compared to what we will often do. They’re called minimum phase filters which means by definition there are already phase errors introduced.
The only way you get no phase errors introduced is by using linear phase crossover filters, and the only way you can do that in a passive design is with a single-pole filter. Using a single capacitor to the tweeter for example and very simplistic electronics, you have almost no protection for the tweeter from bass frequencies, it will just instantly drive it into crazy territory. You need a steeper filter to protect the tweeter from bass distortions, so what we like to do is to suggest to people use pretty steep linear-phase filters which we can do because we are in the digital domain. We are using convolution to do the filters, which means we can maintain linear phase even with very steep filters like 60dB per octave for example, which is a 10-pole filter. And that’s like a brickwall protection to a tweeter, it will shut the midrange driver crossing over to the tweeter. So with a steep crossover, that midrange driver is going to sound more natural because it won’t start beaming. The tweeter will be completely omni-directional, and the midrange will probably still be largely omni-directional, it will give you a much more natural dispersion of frequencies at that crossover region and it just sounds more natural basically.
JK: In your experience so far, you’ve probably measured a couple of different speakers already. Which speakers have you found to have the lowest distortion?
KR: (Chuckles) Look that’s a good question and it’s very sad by the way, that speaker manufacturers do not publish their distortion. They know how bad it is, when people are used to reading distortions of 0.001% with amplifiers, and their speakers are 1% distortion.
Almost all affordable consumer speakers are about 1% distortion. Now I know all about 1% distortion because the Fairlight had that because it only had 8-bits of audio and that was the best it could do. When you have 1% of distortion, you’re getting about 40dB of resolution. In other words, once I get down to -40dB, I start hearing introduced harmonics that shouldn’t be there. In the case of a speaker, fortunately they tend to be the second order harmonics which isn’t too agonizing to listen to, but it does mask resolution of lower detail. Not far behind, is the third harmonic which is really irritating, and you do start noticing it. So what’s great are speakers that can manage about 0.1% distortion and that means they have 60dB of clean resolution before introduced harmonics start arriving.
An example of that would be the magnesium midrange cones and I’m really focusing on midrange because that’s where it’s critical to try to get the distortion as low as you can. Athough tweeters manage to get to 0.1% and that’s why I suspect a lot of people are into these freakishly high sample rates, because the only way they can hear any difference is if the tweeter has enough high resolution to even make that noticeable and tweeters do tend to have quite good resolution above 2kHz. but getting back to the midrange, the SEAS magnesium drivers for example I think they are the Excel range in the 5-inch, they were a popular one that we used to use in a speaker, I quite like the new Purifi drivers that look as ugly as hell, but all of them without exception can manage 0.1% distortion and it’s just astonishing how they do it, I don’t know how they do it. Another one that does 0.1% distortion are the ceramic drivers that cost an absolute arm and a leg.
Above: An example of the SEAS magnesium cone woofer
JK: When you’re talking about distortion, the cabinet is going to play a role too isn’t it?
KR: Absolutely yes.
JK: The drivers, you can measure them directly if it was an open baffle design, but if you were to put them into a cabinet, then the cabinet would potentially contribute to more distortion?
KR: The cabinet can make things worse quite easily. That’s another issue with porting, you’ve got a port to extend your bass, I’m sorry but you’ve got some sounds delayed by its cycle coming out the wrong end of the speaker and that’s not going to do great things to your resolution. It’s not disastrous though. The speakers we had at the SAC meeting had ports in them, they sounded pretty good but their bass extension was just extraordinary, as you heard them.
JK: The little bookshelves (Serhan Swift mμ2), everyone raves about how good it extension is for its size
KR: They’re really excellent, I think Brad did a fantastic job with that crossover, I’m not sure which drivers he’s using.
JK: Scanspeak I think?
KR: Yes, Scanspeak. They’re good, I’m not sure what their distortion is, I haven’t seen their distortion measurements. What I want to do with DEQX is to put in a distortion measurement software so that people can see what the distortion actually is when they’re doing the speaker measurements. That would be useful.
JK: I’d love to see the database of different speaker models. Some speaker companies like Wilson Audio likes to build separate cabinets for each driver. The cabinets can be tilted, or their relative position to each other can be changed for time alignment purposes. What are your thoughts on the end result doing it physically vs doing it via software correction like DEQX
KR: It just shows you how alarming it is, that everyone doesn’t do that. What it tells you, is that absolutely no one except for instance Wilson Audio, probably has an accurate time alignment between their tweeter and their midrange and their woofers because they’re on a flat surface. You can do tricks with the crossover to help to get them time aligned, but Wilson Audio does it physically by pulling it back and you’ll notice that they pull it back in the order of probably 5 or 6cm. The tweeter would be behind the midrange, and that is the sort of timing difference that we’re talking about that can make a difference.
JK: So is it effective in your opinion?
KR: Absolutely it’s effective. The easy way to do it, is to delay the signal but you can only do that if you’re active and you’re digitally active. That’s the trick to a system like the Pre-8 for example, in fact any digital crossover would do that.
JK: But what if you have a passive speaker, and now you connect it using the DEQX preamp, you can delay it right?
KR: That’s true, it will attempt to clean that up. But there will be an area where it can’t unscramble the eggs. But I’m surprised at how well that actually seems to work. It certainly does help. I might also add, ironically it’s harder to measure a full range speaker than to measure an active speaker unless it’s bi-wirable where you can drive the bass woofer with a separate input. It's good if you can get the microphone as close to the driver when you’re measuring it. If you’re measuring your passive speaker, you can only put your microphone in one position. You need to put the mic in between the tweeter and the midrange.
The other thing we can do is to take generic measurements for a model of speakers. Take a speaker like the Wilson Audio Sasha V for example. They’ll be a pain to measure because they’re already a good speaker so you have to get a really good measurement to make them better. That means you don’t really want to be doing this in a small room. It would be good to do this outside where you don’t get reflections from the ground too quickly, so people can spend time getting good anechoic measurements and they can be stored in the cloud. This is not being done yet, but it is on the agenda. But if you do buy a Sasha V speaker, you can look it up and DEQX may already have a filter for it that you can download. It would avoid the need to do any measurements for the speaker yourself. You just download it like you’re getting a printer driver. We will do this for certain models but there are thousands of models out there so it’s impossible for us to provide a filter for each speaker model.
JK: Now I know you’re busy trying to get DEQX through the beta phase at the moment, before launching the products worldwide. During this time, are you 100% focused on DEQX only or does your mind wander to any other ideas, future projects, new products, new ideas for example?
KR: No, not outside of DEQX. We’ve got a lot of projects in DEQX, we’ve got ideas for. I try to stay focused, not too many things at once. We do want to get the multi-channel amplifier out because we feel it’s something people would like. It just makes life easier if we can provide a family-type solution that just plugs together and looks the part. It’s also a great amplifier the Purifi. So that’s the next focus and then we do have surround on the agenda, probably as an add-on to the 2-channel systems. There’s a lot of stuff that’s actually on the longer term thing but I can’t really talk about that. I’m getting too old to have too many things.
JK: Do you currently have a 2-channel audio system running at home and if you do, what are you currently running?
KR: Yeah I do, but I’m sort of like the plumber whose house plumbing never works. My place down at the farm is where we do most of my fiddling around and that is where I am using that SEAS low-distortion driver, and I’ve got a variety of amplifiers that I play around with down there. It’s where I’m able to compare things. I was using beryllium tweeters, I’ve gone back to soft-dome tweeters, I’ve got a couple of 10-inch subwoofers on each side, back-to-back which gives you nice clean bass up to about 200Hz where I’m crossing over to the SEAS midrange, and that’s sort of the system that I tend to listen to. Nothing particularly dramatic.
JK: And which amplifiers have you been using?
KR: I’ve been playing with the Hypex’s originally, I’ve got an AB amp from Germany, I just can’t remember the name of it but it’s 1000W per channel for the bass and the Hypex amps are ok, they’re good but when I put in the Purifi amps, this is sort of magic really. It’s quite a surprise for a Class D amp.
JK: And what’s your source mainly, digital?
KR: Yes, I tend to just stream Tidal, Qobuz, etc.
JK: Thank you for your time Kim. It’s been a very enjoyable chat, and a very informative one too at that.
If you’re interested in improving the performance of your speakers and system with a DEQX product, for a limited time only you can save a significant amount by joining its beta program. Click the link below for more details on how you can join the program and save a significant amount off the retail price.
DEQX Beta Program: https://www.deqx.com/deqx-gen-4-beta-order/
JK: When you’re talking about distortion, the cabinet is going to play a role too isn’t it?
KR: Absolutely yes.
JK: The drivers, you can measure them directly if it was an open baffle design, but if you were to put them into a cabinet, then the cabinet would potentially contribute to more distortion?
KR: The cabinet can make things worse quite easily. That’s another issue with porting, you’ve got a port to extend your bass, I’m sorry but you’ve got some sounds delayed by its cycle coming out the wrong end of the speaker and that’s not going to do great things to your resolution. It’s not disastrous though. The speakers we had at the SAC meeting had ports in them, they sounded pretty good but their bass extension was just extraordinary, as you heard them.
JK: The little bookshelves (Serhan Swift mμ2), everyone raves about how good it extension is for its size
KR: They’re really excellent, I think Brad did a fantastic job with that crossover, I’m not sure which drivers he’s using.
JK: Scanspeak I think?
KR: Yes, Scanspeak. They’re good, I’m not sure what their distortion is, I haven’t seen their distortion measurements. What I want to do with DEQX is to put in a distortion measurement software so that people can see what the distortion actually is when they’re doing the speaker measurements. That would be useful.
JK: I’d love to see the database of different speaker models. Some speaker companies like Wilson Audio likes to build separate cabinets for each driver. The cabinets can be tilted, or their relative position to each other can be changed for time alignment purposes. What are your thoughts on the end result doing it physically vs doing it via software correction like DEQX
KR: It just shows you how alarming it is, that everyone doesn’t do that. What it tells you, is that absolutely no one except for instance Wilson Audio, probably has an accurate time alignment between their tweeter and their midrange and their woofers because they’re on a flat surface. You can do tricks with the crossover to help to get them time aligned, but Wilson Audio does it physically by pulling it back and you’ll notice that they pull it back in the order of probably 5 or 6cm. The tweeter would be behind the midrange, and that is the sort of timing difference that we’re talking about that can make a difference.
JK: So is it effective in your opinion?
KR: Absolutely it’s effective. The easy way to do it, is to delay the signal but you can only do that if you’re active and you’re digitally active. That’s the trick to a system like the Pre-8 for example, in fact any digital crossover would do that.
JK: But what if you have a passive speaker, and now you connect it using the DEQX preamp, you can delay it right?
KR: That’s true, it will attempt to clean that up. But there will be an area where it can’t unscramble the eggs. But I’m surprised at how well that actually seems to work. It certainly does help. I might also add, ironically it’s harder to measure a full range speaker than to measure an active speaker unless it’s bi-wirable where you can drive the bass woofer with a separate input. It's good if you can get the microphone as close to the driver when you’re measuring it. If you’re measuring your passive speaker, you can only put your microphone in one position. You need to put the mic in between the tweeter and the midrange.
The other thing we can do is to take generic measurements for a model of speakers. Take a speaker like the Wilson Audio Sasha V for example. They’ll be a pain to measure because they’re already a good speaker so you have to get a really good measurement to make them better. That means you don’t really want to be doing this in a small room. It would be good to do this outside where you don’t get reflections from the ground too quickly, so people can spend time getting good anechoic measurements and they can be stored in the cloud. This is not being done yet, but it is on the agenda. But if you do buy a Sasha V speaker, you can look it up and DEQX may already have a filter for it that you can download. It would avoid the need to do any measurements for the speaker yourself. You just download it like you’re getting a printer driver. We will do this for certain models but there are thousands of models out there so it’s impossible for us to provide a filter for each speaker model.
JK: Now I know you’re busy trying to get DEQX through the beta phase at the moment, before launching the products worldwide. During this time, are you 100% focused on DEQX only or does your mind wander to any other ideas, future projects, new products, new ideas for example?
KR: No, not outside of DEQX. We’ve got a lot of projects in DEQX, we’ve got ideas for. I try to stay focused, not too many things at once. We do want to get the multi-channel amplifier out because we feel it’s something people would like. It just makes life easier if we can provide a family-type solution that just plugs together and looks the part. It’s also a great amplifier the Purifi. So that’s the next focus and then we do have surround on the agenda, probably as an add-on to the 2-channel systems. There’s a lot of stuff that’s actually on the longer term thing but I can’t really talk about that. I’m getting too old to have too many things.
JK: Do you currently have a 2-channel audio system running at home and if you do, what are you currently running?
KR: Yeah I do, but I’m sort of like the plumber whose house plumbing never works. My place down at the farm is where we do most of my fiddling around and that is where I am using that SEAS low-distortion driver, and I’ve got a variety of amplifiers that I play around with down there. It’s where I’m able to compare things. I was using beryllium tweeters, I’ve gone back to soft-dome tweeters, I’ve got a couple of 10-inch subwoofers on each side, back-to-back which gives you nice clean bass up to about 200Hz where I’m crossing over to the SEAS midrange, and that’s sort of the system that I tend to listen to. Nothing particularly dramatic.
JK: And which amplifiers have you been using?
KR: I’ve been playing with the Hypex’s originally, I’ve got an AB amp from Germany, I just can’t remember the name of it but it’s 1000W per channel for the bass and the Hypex amps are ok, they’re good but when I put in the Purifi amps, this is sort of magic really. It’s quite a surprise for a Class D amp.
JK: And what’s your source mainly, digital?
KR: Yes, I tend to just stream Tidal, Qobuz, etc.
JK: Thank you for your time Kim. It’s been a very enjoyable chat, and a very informative one too at that.
If you’re interested in improving the performance of your speakers and system with a DEQX product, for a limited time only you can save a significant amount by joining its beta program. Click the link below for more details on how you can join the program and save a significant amount off the retail price.
DEQX Beta Program: https://www.deqx.com/deqx-gen-4-beta-order/