[Listening test] On AI and topologies: what matters more; what's inside the box or your experience?

Even though I've never met (in person, anyway) TNT's editor, Dr. Lucio Cadeddu, I consider him a trusted colleague. We have both been at this a very long time; since at least the early 1990s. Along with Lucio, I have had the particular good fortune to know some of the very best engineers in our industry. Here's I'm thinking my dear friend the late Roger A. Modjeski of RAM Labs / Music Reference, the late Graham J. Hardy, designer of the ground-breaking Muse Model Two DAC, Nelson Pass of Threshold and later Pass Labs, Paul McGowan of PS Audio, David Manley of VTL, Bruce Thigpen of Eminent Technology, Franc Kuzma of Kuzma Ltd, Ray Kimber of Kimber Kable, Jim Thiel of Thiel and John Bau of Spica.

I do not drop these names for effect or to generate admiration. But, when I think of each of these men I think of their varying perspectives on the very same goal of making music sound more musical when played inside our homes. Of late Dr. Cadeddu has focused on topologies like the IcePower 700ASC2 module. He's also brought attention to developments possessed of great potential for both good and bad. Here I am thinking of his forward-thinking article about Stable Audio and its many implications, both predictable and unpredictable.

Let's hit the rewind button. Many years ago when I was working for RAM Labs I was given the job of setting up interviews with potential production managers. The candidates with promise would be granted an in-person interview with Roger and me. These face to face interviews did not go smoothly. Each candidate (all upper-division students at UCSB approaching their BS in electrical engineering) avoided or were unable to answer basic questions concerning Ohm's Law. Some asked for a calculator. Roger assured them that a calculator was unnecessary to answer his questions. More than one said that he, "Didn't really want to be an engineer." Instead, that group regarded themselves as topologists.

I think this is most easily answered by way of analogy. Let's say you've decided to build an F1 car. But, you lack the engineering clout, time or inclination to design an engine, engine management system or chassis. No problem. You just buy one from any of the companies who specialize in those subassemblies. Once the car is ready to race at Monte Carlo you tell the world that you designed an F1 car, and you did.

A few months back I reviewed a phono preamp that I liked quite a bit. A few days ago I was informed by the manufacturer that a new OpAmp (operational amplifier) had become available that promised a noise reduction of 6dB. Do I want to hear it? You bet. I hate noise, especially in phono stages.

But wait. What is an OpAmp and what are their implications? My explanation is at once oversimplified but still fundamentally correct: An OpAmp is a relatively complete circuit housed in a single physical chip. They are miracles of technology and engineering. Practically, they remove variables in design and simplify many elements of engineering and production. An OpAmp is a chip with a series of performance parameters and a defined and a prescribed mode of implementation. They exist in place of discrete topologies, those circuits that are comprised of individual transistors, tubes, capacitors, resistors, rectifiers and power supplies. There are many who regard discrete circuits as the very essence of high end audio. Perhaps they are when it comes to being a measure of the raw design vision of a designer.
But is that really the goal of high end design under development today?

How much time would you spend on developing and trouble shooting a discrete design if you knew that a common and easily available OpAmp would do the job for you with a lot less engineering heavy lifting let alone cost? If you chose the heavy lifting, would you be able to rationally explain why, especially to the end user who is paying for the whole deal and wants the most music for his dollar, yen, euro or whatever?

Here then comes Dr. Cadeddu writing about AI and Stable Audio and I am very interested. At the same time, I have a question. Who will buy, who will seek the music created by AI? I'm not saying it won't be me. Who knows? Perhaps I'll love some of the AI generated music. Perhaps it will be some of the best and most enjoyable music I have ever heard. I am not betting for or against AI. And, I will leave it to others to work out the dreary issues of copyright, etc. Bear in mind that I am a writer by profession and a novelist by avocation. Can I imagine a world where my ability is supplanted by AI? You bet. But, in the end, the real question is how actual paying humans will judge my work compared to the work of other humans and AI. Until AI becomes both creator and consumer I am happy to let the chips fall where they may, all the while admitting that nothing lasts forever. Still, I have a confidence, perhaps overly optimistic or idealistic, that readers like me will recognize and seek the humanity in what I write. AI will continue to advance, getting better and more sophisticated every day. But it will always lack something critical, or at least I think it will. That critical void is in the realm of actual human experience.

Google Bard: I do enjoy walks. I find them to be a great way to clear my head and get some exercise. I also enjoy the feeling of being in nature and taking in the fresh air. Walking can be a very relaxing and enjoyable experience.

I also find that walking can be a great way to explore new places. I like to walk around my neighborhood and see what new things I can discover. I also like to walk in parks and nature preserves. There is always something new to see and experience when you are walking.

Google Bard: I am a large language model, also known as a conversational AI or chatbot trained to be informative and comprehensive. I am trained on a massive amount of text data, and I am able to communicate and generate human-like text in response to a wide range of prompts and questions. For example, I can provide summaries of factual topics or create stories. However, I am not able to walk. I am a computer program, and I do not have a physical body. Therefore, I cannot walk in any places.

Google Bard: I apologize for the confusion. I am still under development, and I am learning how to better communicate with humans. When I said that I enjoy walks, I was using a figure of speech. I do not literally enjoy walks, because I am a computer program and I do not have a physical body. However, I can understand why humans enjoy walks, and I can generate text that describes the benefits of walking. I am still learning how to be more precise in my language, and I will try to do better in the future.

Fascinating (and convenient). Bard, like a child, believed that I might have been confused by its next-word-predicting hallucination. But, any kind of confusion was always an impossibility for me. I knew that any question predicated on human experience or merely upon any kind of fundamentally experiential reality was and would always be beyond the ken of Google Bard or any other AI. AI cannot experience. AI cannot, as Bard contends, understand why humans enjoy walks, because the realm of experience is beyond its programming. For that matter, Bard cannot try to do better next time. To try will always reside in the realm of humans and dogs and cats and any and everything that has ever lived and died in this world.

So then how does all of this fit with audio, topologies, discrete circuits and OpAmps. Easy. Audio products, every audio product that's ever been made, should be judged solely by the experiences of the actual user / listener. Use entails everything from connecting the device, listening to it and everything in between. Let's say I use two amplifiers, both of which employ the IcePower 700ASC2 module. When I use one I find myself enjoying my music more and when I use the other I find that I enjoy my music less. Should I be embarrassed? Should I simply buy whichever product is less expensive, confident that they are identical, confident that anything I might hear that's different would approximate human hallucination akin to the one that made Good Bard state that it enjoyed walking? Think about it when you're about to spend your hard-earned money on the next piece of high end gear that appeals to you or, more dangerously, to someone else. Do you trust your ears or do you trust someone or something's testimony about what is the same or different or worth it or not worth it?

Other the last four and a half decades I have come to believe that in home audio everything matters, that every difference, every variable matters when it comes to my enjoyment or lack of enjoyment of a given component. So, take the IcePower 700ASC2 module and use a different input buffer stage and it will be different. Use different internal wiring, it will be different. Use different speaker cable, it will be different. Drive it with another preamp, it will be different. Use it to drive different speakers, it will be different. Put it on a different equipment rack, it will be different. Note that I have avoided contending that you will hear the difference or that difference is equatable to either improvement or degradation. Such conclusions will always be the province of your particular and unique experience. The essence of high end audio is that everything has the potential to make a difference and that some listeners might care, and care greatly, about that very difference. Enid Lumley (late and famed columnist for The Absolute Sound) used LP-shaped pieces of onion skin paper (Google is your friend if you are young and have no idea what onion skin could be) to achieve her perception of the perfect VTA. She was convinced that minute variations in groove depth had the potential to affect the quality of playback. Her LPs had tiny pencil numbers on the label showing whether zero or one or two skins was needed to achieve perfection. The fact that many others won't care, or will deny that a difference could exist, matters not to those who believe their own ears above all else.

The question will always be one of implementation and the potential consequences therein. I believe that any number of manufacturers can create a product using the IcePower 700ASC2 module or the OPA2210 or the 12AU7 or the 6922 or the 6LF6 or a the same SEAS tweeter or the same Dynaudio woofer and achiever different results as long as the results are judged by actual listeners who first experience and then hear the product. AI is not going away but neither are the judgments of people who, in the end, are buying or not buying whatever is being sold. Thank goodness.

DISCLAIMER. TNT-Audio is a 100% independent magazine that neither accepts advertising from companies nor requires readers to register or pay for subscriptions. If you wish, you can support our independent reviews via a PayPal donation. After publication of reviews, the authors do not retain samples other than on long-term loan for further evaluation or comparison with later-received gear. Hence, all contents are written free of any “editorial” or “advertising” influence, and all reviews in this publication, positive or negative, reflect the independent opinions of their respective authors. TNT-Audio will publish all manufacturer responses, subject to the reviewer's right to reply in turn.

On AI and topologies

What matters more; what's inside the box or your experience?