Taiko Audio SGM Extreme : the Crème de la Crème

Interesting how every generation views a product as “perfect” only for that opinion to change as time goes by. Early phonograph records were considered perfect and better than live music. Folks with long memories may recall when compact discs were first introduced in the early to mid 1980s, they were advertised as “perfect sound forever”. For many years, each CD had some boilerplate text at the back claiming that CDs were best sound one could obtain etc. Now of course everyone thinks Roon streaming or some variants are perfect.

I’m holding out hope for a better recording technology in the decade to come. To me PCM is fundamentally flawed and no amount of skullduggery with transports or DACs will fix the inherent flaws of PCM technology. Unlike analog, in PCM, distortion greatly rises as the volume reduces, so that quiet passages — say a single oboe playing in an orchestra — are rendered with very poor fidelity. Unlike analog noise, which is largely uncorrelated with with signal, digital PCM noise is correlated with signal intensity.

Human hearing is inherently nonlinear and highly adaptive to signal level. A much better technology than PCM would adaptively allocate bits to where human hearing is acutely sensitive — the midrange and in particular quiet to normal volumes — and not waste it in regions where we are mostly deaf or in having dynamic range beyond what any listening room can possibly support (e.g., above 20 kHz and below 30 Hz, and dynamic range above 80-90 dB).

There have been a lot of theoretical advances in signal processing since the age old Nyquist theorem that underlies PCM technology (i.e., Nyquist theorem mandates sampling at least twice the rate of the highest frequency, so CD technology used 44.1 kHz as the original baseline). Some of these advances like compressed sensing have transformed other areas, but not yet been applied to music, which continues to be based on largely obsolete math. But that will change in the coming decades, and PCM will become a quaint historical artifact.

The real revolution will occur when quantum computing takes over and we can reliably transport qubits using quantum entanglement (which won this year’s Nobel prize in physics). That will make today’s internet look like chiseling on a stone tablet, like the Babylonians did 5000 years ago. Qubit encoded music should be able to provide far higher fidelity than any PCM encoding can achieve.

Working in that field, I can confirm we have a good 30 years to sit happy with our Extreme until that happens :) Fascinating field however!
 
Working in that field, I can confirm we have a good 30 years to sit happy with our Extreme until that happens :) Fascinating field however!
Yes, I expect you’re right. But who knows? Perhaps things will move more quickly. After DSD encoding, the search for alternative codecs seems to have died. I wish more effort is put into this neglected area. It’s like digital cameras. No matter how many megapixels you have in your CCD array, all digital cameras only see grayscale, not color. You need the Bayer color array filter to guesstimate color, which is why color reproduction on digital cameras is so awful. Once you shoot with the Leica Monochrom, which gets rid of the color array filter and gives you the pure CCD output, it’s shocking to see how good that pure grayscale image looks. For a true advance in digital photography we have to wait till someone figures out how to invent a pure color CCD array. Similarly digital reproduction is waiting for a better solution than PCM.
 
Yes, I expect you’re right. But who knows? Perhaps things will move more quickly. After DSD encoding, the search for alternative codecs seems to have died. I wish more effort is put into this neglected area. It’s like digital cameras. No matter how many megapixels you have in your CCD array, all digital cameras only see grayscale, not color. You need the Bayer color array filter to guesstimate color, which is why color reproduction on digital cameras is so awful. Once you shoot with the Leica Monochrom, which gets rid of the color array filter and gives you the pure CCD output, it’s shocking to see how good that pure grayscale image looks. For a true advance in digital photography we have to wait till someone figures out how to invent a pure color CCD array. Similarly digital reproduction is waiting for a better solution than PCM.
I fully agree and really interesting! Just in case of int, you might find the Foveon sensor (now bought by Sigma corp) intriguing, it bypasses the use of Bayer filter by stacking photodiodes: https://en.m.wikipedia.org/wiki/Foveon_X3_sensor
 
Interesting how every generation views a product as “perfect” only for that opinion to change as time goes by. Early phonograph records were considered perfect and better than live music. Folks with long memories may recall when compact discs were first introduced in the early to mid 1980s, they were advertised as “perfect sound forever”. For many years, each CD had some boilerplate text at the back claiming that CDs were best sound one could obtain etc. Now of course everyone thinks Roon streaming or some variants are perfect.

I’m holding out hope for a better recording technology in the decade to come. To me PCM is fundamentally flawed and no amount of skullduggery with transports or DACs will fix the inherent flaws of PCM technology. Unlike analog, in PCM, distortion greatly rises as the volume reduces, so that quiet passages — say a single oboe playing in an orchestra — are rendered with very poor fidelity. Unlike analog noise, which is largely uncorrelated with with signal, digital PCM noise is correlated with signal intensity.

Human hearing is inherently nonlinear and highly adaptive to signal level. A much better technology than PCM would adaptively allocate bits to where human hearing is acutely sensitive — the midrange and in particular quiet to normal volumes — and not waste it in regions where we are mostly deaf or in having dynamic range beyond what any listening room can possibly support (e.g., above 20 kHz and below 30 Hz, and dynamic range above 80-90 dB).

There have been a lot of theoretical advances in signal processing since the age old Nyquist theorem that underlies PCM technology (i.e., Nyquist theorem mandates sampling at least twice the rate of the highest frequency, so CD technology used 44.1 kHz as the original baseline). Some of these advances like compressed sensing have transformed other areas, but not yet been applied to music, which continues to be based on largely obsolete math. But that will change in the coming decades, and PCM will become a quaint historical artifact.

The real revolution will occur when quantum computing takes over and we can reliably transport qubits using quantum entanglement (which won this year’s Nobel prize in physics). That will make today’s internet look like chiseling on a stone tablet, like the Babylonians did 5000 years ago. Qubit encoded music should be able to provide far higher fidelity than any PCM encoding can achieve.


It seems to me you are mainly addressing myths that have been properly explained and debunked since long about PCM - but it is not necessary to go again on this discussions, as current top PCM, (24 bits at 352 KHz) has none of these problems and fortunately the maths behind top modern recording systems takes in consideration research in psychoacoustics and are no way obsolete.

IMO the real limitations in audiophile sound reproduction are the stereo standard and the listeners, not the current limitations of computing and storage.

The same way there is no "best sound" there is no possibility of having perfection in sound reproduction - even if we could achieve a perfect physical sound field facsimile, this is not the objective of sound reproduction. So, it is natural that each large step is considered the perfection at the moment. And again just my opinion, we have had great and enjoyable CD sound for decades - I would say since the middle 90's. Surely YMMV.
 
It seems to me you are mainly addressing myths that have been properly explained and debunked since long about PCM - but it is not necessary to go again on this discussions, as current top PCM, (24 bits at 352 KHz) has none of these problems and fortunately the maths behind top modern recording systems takes in consideration research in psychoacoustics and are no way obsolete.

IMO the real limitations in audiophile sound reproduction are the stereo standard and the listeners, not the current limitations of computing and storage.

The same way there is no "best sound" there is no possibility of having perfection in sound reproduction - even if we could achieve a perfect physical sound field facsimile, this is not the objective of sound reproduction. So, it is natural that each large step is considered the perfection at the moment. And again just my opinion, we have had great and enjoyable CD sound for decades - I would say since the middle 90's. Surely YMMV.
Let’s take your choice of 24-bit 352khz PCM. Leaving aside the problem that 99.99% of all digital recordings are still 16 bit 44.1 kHZ (sampling from Tidal or Qobuz will make this obvious), what have we gained from this huge increase in file size? It’s not clear to me that expending enormous resources reproducing frequencies like 100 khz is useful since there’s no audio content there whatsoever. Further, it would be nice if we had DACs that can truly resolve 24-bit depth. Do you know any? I certainly don’t. Note just because a DAC designer uses a DAC chip that takes 24-bit input means zilch. John Atkinson in Stereophile has been measuring DACs for 25+ years. The best DACs can barely resolve 20 bits. Read his detailed measurements. Many other reviews in ASR and other places confirm this problem. It’s very hard to get to true 24-bit linearity. Noise has to be less than -140 dB. In the many years I have owned audio equipment, only one component has come close, the pure analog Mark Levinson 32 preamp, which uses a sophisticated power supply that converted AC line voltage to DC and then regenerated pure AC back again. Measurements in Stereophile of the 32 preamp showed all noise under -130 dB. Few if any DACs get anywhere near that.

For example, the highly raved about Chord M-scaler gets 16-bit of resolution. You pay 6 grand for throwing away 8 bits of resolution from your 24-bit data stream. Isn’t that a laugh?


Umm.. let’s see what I get if I spend 15 grand and get the highly prized Chord Dave. Oops, we get perhaps 18 bits of resolution from this very expensive DAC.

None of this should be interpreted as saying digital audio is not enjoyable or is not a significant step forwards. It has brought much joy into my life these past 35 years. But it’s a far cry from the sound of actual live music in a concert hall.


I could go on and embarrass other DAC manufacturer, but it’s important to realize that bit depth and sampling frequency is a game that’s played to somehow convince that you’re really getting your money’s worth when you’re not. In fact, money spent on a DAC has zero correlation with performance, most of time actually negative correlation. The best performing DACs are often very cheap ones.

Regarding math. Simply expanding bit depth to 24 bits or sampling to 352 kHZ doesn’t change the underlying math at all, just like going to a 100 megapixel camera doesn’t change the fact that CCD sensors only capture gray scale information. If you want to see the game changing math that goes far beyond Nyquist sampling, read about compressive sensing. That’s truly new math that will one day change digital audio.


 
  • Like
Reactions: Nils
Maybe this discussion should be moved to a separate topic since it isn’t unique to Taiko Audio’s Extreme.
 
Agreed. None of this implied anything towards the Taiko Extreme, which is clearly a beautifully engineered piece of hardware. I admire folks who can put in so much time and energy on such a product when its sales volume must be quite small. Hopefully some of its qualities can be distilled down to lower price points. My only point was that there are deeper issues with digital reproduction that Taiko alone will not solve, nor is it intended to. I agree that discussion shouldn’t be here. Apologies!
 
  • Like
Reactions: Nils and Rhapsody
For example, the highly raved about Chord M-scaler gets 16-bit of resolution. You pay 6 grand for throwing away 8 bits of resolution from your 24-bit data stream. Isn’t that a laugh?

Umm.. let’s see what I get if I spend 15 grand and get the highly prized Chord Dave. Oops, we get perhaps 18 bits of resolution from this very expensive DAC.

Excellent argument for not choosing hardware based on that particular measurement (as it was captured) if the goal is great sound quality. It clearly conveys little that’s useful in assessing how good these products actually sound.
 
Excellent argument for not choosing hardware based on that particular measurement (as it was captured) if the goal is great sound quality. It clearly conveys little that’s useful in assessing how good these products actually sound.
Well said.
 
  • Like
Reactions: kennyb123
Agreed. None of this implied anything towards the Taiko Extreme, which is clearly a beautifully engineered piece of hardware. I admire folks who can put in so much time and energy on such a product when its sales volume must be quite small. Hopefully some of its qualities can be distilled down to lower price points. My only point was that there are deeper issues with digital reproduction that Taiko alone will not solve, nor is it intended to. I agree that discussion shouldn’t be here. Apologies!
Nevertheless, this was an interesting read!
 
  • Like
Reactions: MagnusG and wil
I think I speak for both Ed and Emile when I say that we have no intention of having XDMS reviewed while it is in the Alpha state, and probably also not when in the Beta state. My personal stance is that we should only do that once both the backend and the front end are entirely stable and working as intended, and when the team collectively feels that we are offering precisely the sound quality that we are after, along with all the functionality that we envision as being essential.
 
Is there an approximate date, for the jump to the beta version, of the back-end and front-end? I want to remember reading to Emile that he hoped to be able to offer a good leap in quality, in sound, before the end of the year.
 
Is there an approximate date, for the jump to the beta version, of the back-end and front-end? I want to remember reading to Emile that he hoped to be able to offer a good leap in quality, in sound, before the end of the year.
We are moving full-steam ahead but there is no projected date for this. It all depends on how quickly some issues are solved and how many other issues we encounter. Reality has a way of inserting loopholes that interfere with our hopes...;-)
 
We are moving full-steam ahead but there is no projected date for this. It all depends on how quickly some issues are solved and how many other issues we encounter. Reality has a way of inserting loopholes that interfere with our hopes...;-)
I for one am SO EXSTATIC with the sound and interface right now, that to me "what's the hurry". We are getting updates every week with Alpha. BUT IT"S GREAT NOW. But that's just me and for my listening preferences.

It think the team is going as fast as they can and we now have a great product and GREAT SQ, while we we wait for future enhancements.
 
I for one am SO EXSTATIC with the sound and interface right now, that to me "what's the hurry". We are getting updates every week with Alpha. BUT IT"S GREAT NOW. But that's just me and for my listening preferences.

It think the team is going as fast as they can and we now have a great product and GREAT SQ, while we we wait for future enhancements.
I totally agree. If it is SQ you chase, look no further. Between Christiaan, Ed and the Taiko team the platform is evolving weekly with each new build which seem to be coming on a much more regular basis
 
I totally agree. If it is SQ you chase, look no further. Between Christiaan, Ed and the Taiko team the platform is evolving weekly with each new build which seem to be coming on a much more regular basis
From SQ pow - XDMS has since it's inception been a no-brainer, after the backend update v221110 SQ is radically improved.

Feature sets are built out incrementally but XDMS Alpha is very usable already now.

Technical skills required to participate in Alpha has been reduced I would say:
1) Most incremental feature updates come via control-app updates - for iOS these are delivered automagically via an app called TestFlight -> no technical skills required

2) More rarely (most recently v221110) there are backend updates - these require a little more technical skills, but since many/most of Alpha testers do these updates on our own there should be bandwidth for some additional Alpha users even if they need assist by @EuroDriver

Given the stellar SQ, anyone on the verge of jumping into XDMS-pool - my advise is to go ahead...
 
From SQ pow - XDMS has since it's inception been a no-brainer, after the backend update v221110 SQ is radically improved.

Feature sets are built out incrementally but XDMS Alpha is very usable already now.

Technical skills required to participate in Alpha has been reduced I would say:
1) Most incremental feature updates come via control-app updates - for iOS these are delivered automagically via an app called TestFlight -> no technical skills required

2) More rarely (most recently v221110) there are backend updates - these require a little more technical skills, but since many/most of Alpha testers do these updates on our own there should be bandwidth for some additional Alpha users even if they need assist by @EuroDriver

Given the stellar SQ, anyone on the verge of jumping into XDMS-pool - my advise is to go ahead...
Is it still the case that it isn’t possible to switch between Roon and XDMS on the iOS app? Once that functionality is present, I would have less hesitation in asking for the alpha build to be installed.
 
Is it still the case that it isn’t possible to switch between Roon and XDMS on the iOS app? Once that functionality is present, I would have less hesitation in asking for the alpha build to be installed.
No, you can't switch between XDMS and Roon with the Icon yet.
 
Is it still the case that it isn’t possible to switch between Roon and XDMS on the iOS app? Once that functionality is present, I would have less hesitation in asking for the alpha build to be installed.
This is on the list and was actually raised in importance today:) Speaking of which, Search by Composer has also been raised in importance. I know many are eagerly awaiting this. (Sorry, I don't know yet when these functions will be implemented...).
 
Great!!!
 

About us

  • What’s Best Forum is THE forum for high end audio, product reviews, advice and sharing experiences on the best of everything else. This is THE place where audiophiles and audio companies discuss vintage, contemporary and new audio products, music servers, music streamers, computer audio, digital-to-analog converters, turntables, phono stages, cartridges, reel-to-reel tape machines, speakers, headphones and tube and solid-state amplification. Founded in 2010 What’s Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals, we enjoy learning about new things and meeting new people, and participating in spirited debates.

Quick Navigation

User Menu