Tube vs Solid State.... is it mostly illusion?

[LerxstLee]

I think alot in black and white. Some think is shades of grey. I am sure it is an easier way of life thinking in grey.
When it comes to tubes and solid state, it is black and white again for me. I think of tubes in an analogue way.
Yet, when it comes to solid state I think of that as digital.
Listen to a tape, LP, even a cd "not a copy" on a tube stereo at home. Then take out that Ipod thing,
or like some people do, grab their "custom" cd with 500 songs, crushed in with 500 mp3s. Where is the imaging.
Where is the detail. My son hears very little difference, yet I won't even listen to that compressed clicking breathing
ones and zeros trying to mess my mind into thinking it is music. Even the highest rate MP3s are easily identified from the
source. I don't mind some mp3s to learn a song, its quick and simple. Yet to listen to mp3s for enjoyment, not in my lifetime. I tried to keep on track, I strayed a tad but, with tubes I hear the open sound stage. Solid state gets
very narrow.
Simply solid state components are either one or off, a 1 or a zero. Yet with a tube, the information has to flow, for simplicity sakes from a plate thru a grid to a cathode. Yes , not on or off, just flowing.
Many variables. Yet do not take away my tubes .

 

[silentrage]

That was really well said Ando, and I hope more artists / engineers take on the job so we could collectively enjoy tube sound AND feel with solidstate utility!

@Lerx
Comparing LPs to 500 mp3 cds is like comparing a ferrari to a crippled donkey. Would you be so confident at a blind test between a LP and a full quality bluray audio disc?
The real world is 1s and 0s too, those 1s and 0s are just planck scale, once we're able to engineer down to that level, digital becomes analog, but the human brain will fail to recognize the difference long before that happens.

 

[Heritage Softail]

Interesting comparison here. When you eat asparagus it causes a chemical reaction. Oddly enough, some people it makes thier piss smell funny, others not, some people can smell it, and others can't. It has nothing to do with cleaning your kidneys. Very individual reaction to this one food item. I think some people are more sensitive to the linear harmonics of a SS amp and do not like it. Others enjoy the sound. Can't we all just accept people are different, sing Kumbaya around the camp fire, playing thru our Solid State or Tube amps. But if you do have a nice Roadster, in good condition, and don't like that tube sound, I will take it off your hands.

 

[silentrage]

Individual taste is all good by me man, but there is no animosity here, just a friendly discussion about whether digital technology will be capable of succeeding or superceding vacuum tube technology one day.

 

[giorikas81]

Interesting point about the Lp/mp3/bluray audio disc. In any case digital can overcome analog. Why? Because it is endless...In a way the whole point comes down to if it is necessary.

Examples. Digital cameras...Proffessional film (not the common one in kodaks of old). How many megapixels will be required until digital can equal analog film in information capacity? And then clarity...? I remember reading about it. The number was 32 or 64 megapixels? The point was that after 12-18 megapixels most eyes did not comprehend a difference. Other means of examintaion like microscopes can.

Then comes the sound. LP sound was measured and I think that 192 khz was in fact the level of information an LP can store? I find the LP nostalgic sweet lyrical but coloured meaning that it is great enhances the experience but is audibly far from perfect. Then again you may see a usb tommorrow with 10x the LP capacity but the way it stores it may not be the same and as a result...

 

[MrMarkIII]

@ ando:
"... Most tube amp makers are musicians who learned a little electronics... "

Hmmm, well, neither Leo Fender nor Randall Smith could actually play guitar.
Just sayin' ...

 

[tmac]

It begs the philosophical question of whether humans can actually conceive of any machine that rivals its own complexity. Personally, I don't think so. Humans will have to evolve further to produce computers that rival our current brain. Our understanding of the brain is still very primitive - as are our instruments for measuring brain activity. What that means is that neural network scientists are making an awful lot of guesses and premises about how the brain works and then trying to model their computers on these premises. We're not even close - trust me. And I'm good friends with a neural network professor. Even he admits we're just getting started on this stuff.

I wouldn't place a bet that it is possible to prove that humans cannot create something more advanced than their own brains.

Already the human race have made intellectual achievements that are out of reach of any one person's intellect. The entire structure of mathematics and physics is more advanced than anything any one individual is able to understand/learn in a lifetime. This comes about as a result of collaboraton, or "collective intelligence", if you will

Also, the collaboration between one human and one computer is very powerful.

So I believe that it is in principle possible to create artificial intelligence that exceeds our own in every way. If it is going to happen or not, that is another question. It will depend on us not ruining everything for ourselves before that times would come. I myself don't really see it as some kind of ultimate scientific goal, however. Trying to describe the fundamental laws of nature is a much more interesting effort (IMO).


EDIT (I should also say something relevant to the OPs question):

Anything that sounds great is fine by me. I want the construction of my amps to so that I can fix them myself, and easily replace components. I don't want modern circuitboards, because it makes the amp harder to mod. However, I do have a Mesa Express 5:25, because it was really cheap (used, and with trivial malfunction that I fixed), and it is great sounding. I might still try to mod it a bit, despite its circuitboards! E.g. remotely switchable FX loop, and possibly a solo boost.

- Torquil

 

[LerxstLee]

A quick look under the hood of the Mark V might be a look at an amp with tubes. Yet, just because it has
tubes does not make is user service friendly. This might be the a great example so far.
Just a thought.

 

[Anonymous]

If you expect a trully boutique amp builder to be there for you personally to custom tailor an amp to your liking plus its responses through some components, the same thing Cliff from Fractal audio/Axe-fx does. But he does it on the digital domain? Why? Because he can. Others can't. He can. He designed the Axe-fx soleheartedly, the results are spectacular, each model consists of 50000 lines of code, is as custom as any tube amp can get with bright virtual capacitors, Sag switch etc. He is every bit a guitar audiophile as Randall Smith and he succeeded in much more difficult situations and not on a blooming market but on a rather dead one. He is not just a computer engineer. Maybe not a true pioneer but every bit a butique builder for what I'd want from one. His product isn't cheap in fact you pay for it as much as a mark V or mark IIC+ used and still you need a power amp and speakers. But its sound is top notch and as a preamp it matches any marshall or mesa preamp we managed to put side by side for raw signal...

There is a demand for his product, and I can respect that; hats off to him for doing well because it's certainly not easy. There are absolutely innovations in guitar tones being made; Mesa - and other companies - would have no use of new patents otherwise. This isn't debatable. As to the Axe-FX - I think it's a very smart product and quite impressive, and to you it might sound as good, but it doesn't to me sound as "natural" as what I've come to love in a guitar tone.

What I was trying and failed to get across earlier is that with digital, you must be an expert to work on the stuff; an amateur can't approach it with decent results like they can with tube technology. I suppose the tubes will still be around though, so this is a pretty lousy argument on my part.

However, what can't be ignored is the fact that if and when digital gets to the point where most newer musicians - and by that I mean people who aren't too set in their ways with tubes - begin using it, I think it's only a matter of time before the money starts to disappear and go towards the cheapest software available. The new EL84 Mesa amp is $900 because in order for the components to be what they are, it has to be $900, and people are already griping about it. Digital can be sold for next to nothing since there doesn't have to be a physical product if people have a computer and download it. When I said that digital takes the "art" out of amp building, this is what I have in mind: some foreign country low-balling the industry into outsourcing jobs. Yes, it's a bit extreme and all things considered, not likely to happen because it won't be the best product. At the same time, I can't help but think at least some quality and "art" will be diminished. It's entirely possible that I'm wrong; nobody can predict the future!

There are some very, very convincing Axe-FX clips out there. Many sound better to me than boutique recorded tube amps that people have done, probably because the Axe-FX doesn't require micing and can sound its very best with a preset, not subjective mic placement, rooms, preamp, etc. If I heard a song on the radio recorded with the Axe-FX, I'm sure I'd have nothing to complain about and might not even realize the difference. But, at this moment in time, I believe the best guitar sound still comes from select tube amps, and that digital has not yet replicated or improved upon the feel of a great tube amp.

 

[ando]

It begs the philosophical question of whether humans can actually conceive of any machine that rivals its own complexity. Personally, I don't think so. Humans will have to evolve further to produce computers that rival our current brain. Our understanding of the brain is still very primitive - as are our instruments for measuring brain activity. What that means is that neural network scientists are making an awful lot of guesses and premises about how the brain works and then trying to model their computers on these premises. We're not even close - trust me. And I'm good friends with a neural network professor. Even he admits we're just getting started on this stuff.

I wouldn't place a bet that it is possible to prove that humans cannot create something more advanced than their own brains.

Already the human race have made intellectual achievements that are out of reach of any one person's intellect. The entire structure of mathematics and physics is more advanced than anything any one individual is able to understand/learn in a lifetime. This comes about as a result of collaboraton, or "collective intelligence", if you will

Also, the collaboration between one human and one computer is very powerful.

So I believe that it is in principle possible to create artificial intelligence that exceeds our own in every way. If it is going to happen or not, that is another question. It will depend on us not ruining everything for ourselves before that times would come. I myself don't really see it as some kind of ultimate scientific goal, however. Trying to describe the fundamental laws of nature is a much more interesting effort (IMO).


EDIT (I should also say something relevant to the OPs question):

Anything that sounds great is fine by me. I want the construction of my amps to so that I can fix them myself, and easily replace components. I don't want modern circuitboards, because it makes the amp harder to mod. However, I do have a Mesa Express 5:25, because it was really cheap (used, and with trivial malfunction that I fixed), and it is great sounding. I might still try to mod it a bit, despite its circuitboards! E.g. remotely switchable FX loop, and possibly a solo boost.

- Torquil

Interesting thoughts Torquil, I will take you up on something though: yes it is true that no single human has been able to traverse the entire sum of human thought and knowledge, but that's partly because there isn't enough time in a lifetime to cover it all. Different people specialise in so many areas of knowledge that that will always be impossible. Still, it's fair to say that there is somebody in each field who has a pretty good handle on things in that and related fields. There is no reason to assume that a computer will tie together the many areas and strands of knowledge any more successfully than the various institutes, universities, historical societies and individuals who comprise the bulk of human endeavour. In fact, most fields use their own internal lexicons, internal logic and methodologies which makes this harder. So when trying to envision a computer which ties together all knowledge and wisdom, you have to start with a complete reworking of the way all of these fields talk to each other and the terms they use within themselves. That is no small feat. Humans are capable of context learning and discrimination in a way that computers may never do. Of course you can speculate that they will, but at this stage, programming computers to think has been an exercise of very strict parameters and limitations. Once you start demanding that a computer can follow a thread of knowledge that contains millions of inconsistencies and still come out with a cohesive thought, you are talking about something exceedingly advanced. And what about all the debates within all fields of knowledge - within the "collective consciousness"? Do you think that a computer will be able to maintain an integrated self and still be able to keep track of all the debates and disagreements within the various fields of study? And then, what about the emotional dimension? Is emotion and subjectivity one of the things that guides all forms of human endeavour? What happens when that part of it is missing from the computer's perspective? And you can't program that into a computer because it is not even understood by humans themselves - good luck trying to model that.

When you consider all these factors you realise that being human is more than being the sum of many parts - it is an organism with so many facets and many of them indescribable by logic - which computers demand. None of that Turing test nonsense is going to get past that hurdle. The human brain is able to keep functioning even with the knowledge of all its logical and emotional inconsistencies. It is able to dream, to aspire and to love and feel. Many of the unpredictable aspects of humanity have given rise to the inventions and insights that we have. How is your computer which doesn't contain these "imperfection" going to access inspiration and creativity? What about the invention that comes from cultural fighting and war? Computers won't be starting wars with each other. Humans got to where they are through an enormous series of illogical and logical developments. Then there's the spiritual component...need I say more?

Computers are nowhere near that level of sophistication and I personally believe they never will be - at least not in the human sense. Maybe there will come a day when they are self-perpetuating intelligences, but they won't be something you would compare to a human brain. I don't see this as a problem though. We don't actually need machines that rival or exceed us - the idea that we do is just another one of those human logical inconsistencies.

It's ok that computers are a bit dumb because making a nice amp will one day be well within the average computer's grasp.... :wink:

 

[silentrage]

That was very well said, I think you laid out the problems AI faces.

I think instead of a magical transformation that turns dumb, logical based AI into human level intelligence will be a slow process, like puzzles falling into place.
AI is already great at storing data, much better than the human brain in fact, and it's making measurable progress in correlating and analyzing that data, for example: http://www.wired.com/wiredscience/2009/04/newtonai/

We're developing methods of AI design that mimick biological intelligence, things like programs based on genetic algorithms and software that emulate neuron firings and synaptic connections being made as a form of learning.

The human brain is indeed amazingly complex, but I don't buy into the more than sum of our parts idea, I think we just havn't broken ourselves down to the smallest constituents yet.

The human brain might do a thousand things that AI can't do right now, but that list is only shrinking and not growing.

 

[ando]

The human brain is indeed amazingly complex, but I don't buy into the more than sum of our parts idea, I think we just havn't broken ourselves down to the smallest constituents yet.

I agree, I think we are still discovering parts and trying to work out what they do. Until we've done that, AI will be necessarily premature and speculative.