Mind gymnastics

[magiccarpetride]

OK, here is some mental gymnastics for you:

Considering that the only purpose of placing a vinyl disc onto a platter that is spinning at 33/4 rotations per minute is to agitate the cantilever whose motions generate electrical signal inside the body of a cartridge, wouldn't it be better if we'd design a sophisticated robotic hand to instigate such motions?

Instead of the blueprint of the recording being stored inside the microgrooves, we could store the blueprint digitally. After all, the vinyl blueprint is merely telling the cantilever when and how to move. Once we agitate the cantilever to move according to the specs, its motion will result in electrical signal which is then transmitted to the phono preamp etc. etc.

The advantage of this hypothetical solution is that we would completely eliminate the relatively brittle signal carrier -- vinyl disc. No wear and tear anymore. And the robotic hand movement will be analog, thus avoiding issues with digital conversion (DAC).

Best of both worlds? But is it doable? Is there an engineering hero who can walk on water and part the red sea and make such an amazing little robot?

[Firebottle]

Seems like what you are talking about is a 'reverse cartridge', this already exists in the shape of a vinyl cutting head.

The problem would be coupling the two together. However with the limitations of the vinyl medium out of the way you could possibly dispense with the need for RIAA correction and excite the cartridge with a flat response.

[walpurgis]

Where is the information stored that would agitate this robotic hand?

[Macca]

Where is the information stored that would agitate this robotic hand?

Digitally, according to the o/p.

Which begs some questions...

[walpurgis]

The awkwardness of round flat records could be avoided, so could the complexity of two channels requiring lateral groove modulations. Simply modulating vertically for one signal and keeping the info on the exterior of a rotating cylinder instead of a disc would ease the situation.

I believe work has been progessing to evolve such an advanced design.

http://i68.tinypic.com/2n8treu.jpg

[Firebottle]

Eliminating the vinyl disc sounds appealing. Then the device which traces the grooves - the stylus - could be eliminated too. The idea is so appealing that Philips and Sony invented the CD player. Since then, the optical disc and laser have been eliminated too.

Thoughts that went through my mind as well :mental:

[magiccarpetride]

The awkwardness of round flat records could be avoided, so could the complexity of two channels requiring lateral groove modulations. Simply modulating vertically for one signal and keeping the info on the exterior of a rotating cylinder instead of a disc would ease the situation.

I believe work has been progessing to evolve such an advanced design.

http://i68.tinypic.com/2n8treu.jpg

Back to the future!

[magiccarpetride]

Eliminating the vinyl disc sounds appealing. Then the device which traces the grooves - the stylus - could be eliminated too. The idea is so appealing that Philips and Sony invented the CD player. Since then, the optical disc and laser have been eliminated too.

Part of my daily job is related to machine learning. Given some messy entangled data set, it is apparently possible to let the machine at it, and sit back and watch what is the machine going to 'learn' from the alphabet soup you're feeding it. We're talking huge quantities of data, something that even the most intelligent humans cannot possibly grok.

To me, recorded music is this gigantic cloud of 'alphabet soup', a huge mass of illegible information, the somehow gets distilled into physical microgrooves. Or sampled into a digital file.

But the sampled file is just the minuscule tip of the iceberg, and the vast majority of the 'alphabet soup' gets willfully ignored in the process of sampling. But with the new machine learning technology, maybe we are getting to the point where the entire 'alphabet soup' gets ingested and fully digested by the machines. In that case, a machine could actually fully reverse-engineer EVERYTHING that's been stored in the microgrooves.

At that point, the information that's been successfully processed by the machine could be used to drive the super-precision tiny robotic hand which will vibrate and generate electrical signal. That would be vastly different from the laser beam reading dots on a spinning disc.

Daydreaming on a cloudy Wednesday morning...

[Macca]

Part of my daily job is related to machine learning. Given some messy entangled data set, it is apparently possible to let the machine at it, and sit back and watch what is the machine going to 'learn' from the alphabet soup you're feeding it. We're talking huge quantities of data, something that even the most intelligent humans cannot possibly grok.

To me, recorded music is this gigantic cloud of 'alphabet soup', a huge mass of illegible information, the somehow gets distilled into physical microgrooves. Or sampled into a digital file.

But the sampled file is just the minuscule tip of the iceberg, and the vast majority of the 'alphabet soup' gets willfully ignored in the process of sampling. But with the new machine learning technology, maybe we are getting to the point where the entire 'alphabet soup' gets ingested and fully digested by the machines. In that case, a machine could actually fully reverse-engineer EVERYTHING that's been stored in the microgrooves.

At that point, the information that's been successfully processed by the machine could be used to drive the super-precision tiny robotic hand which will vibrate and generate electrical signal. That would be vastly different from the laser beam reading dots on a spinning disc.

Daydreaming on a cloudy Wednesday morning...

All your theories seem to be based on a basic misunderstanding of what a 'music signal' actually is and a complete lack of knowledge as to how it is reproduced. Sorry if that sounds a bit harsh, it isn't meant to be. A sound recording is a very simple set of data. That's why Edison's cylinder worked.

[magiccarpetride]

All your theories seem to be based on a basic misunderstanding of what a 'music signal' actually is and a complete lack of knowledge as to how it is reproduced. Sorry if that sounds a bit harsh, it isn't meant to be. A sound recording is a very simple set of data. That's why Edison's cylinder worked.

Interesting. But if it is indeed that simple, why so much discrepancy between different players? You'd expect that a simple thing produces predictable, easily reproducible output?

[Macca]

Interesting. But if it is indeed that simple, why so much discrepancy between different players? You'd expect that a simple thing produces predictable, easily reproducible output?

Voicing- people don't want neutral, on the whole - and standard of engineering. The signal may be simple but it isn't difficult to ruin it on its way from storage medium to sound pressure waves.

[magiccarpetride]

Voicing- people don't want neutral, on the whole - and standard of engineering. The signal may be simple but it isn't difficult to ruin it on its way from storage medium to sound pressure waves.

So what you're saying is that different players are purposefully designed to alter the simple signal?

[Macca]

So what you're saying is that different players are purposefully designed to alter the simple signal?

Yes of course they are.

[magiccarpetride]

Yes of course they are.

Why would they want to spoil the broth? It's like buying a TV that overlays colours over the original movies. That's crap.

[Macca]

It's more like slightly tweaking the colour or the brightness on your tv, to your individual taste.

There's a difference between presentation and quality. If you take CD players, they are mostly voiced one way or another. They also differ widely on how well they are engineered where it matters (power supply and the analogue output stage). The voicing determines the presentation, the engineering determines the quality of the sound.

So even if you prefer the voicing of Marantz to say, Sony, your still going to acknowledge that a heavily engineered Sony sounds better than a cost cut-engineered Marantz player. I mean I prefer the Sony voicing but I'd still take a flagship Marantz over a budget Sony any day.

I think conflating presentation and quality leads to a lot of needless arguments on forums.

[magiccarpetride]

It's more like slightly tweaking the colour or the brightness on your tv, to your individual taste.

There's a difference between presentation and quality. If you take CD players, they are mostly voiced one way or another. They also differ widely on how well they are engineered where it matters (power supply and the analogue output stage). The voicing determines the presentation, the engineering determines the quality of the sound.

So even if you prefer the voicing of Marantz to say, Sony, your still going to acknowledge that a heavily engineered Sony sounds better than a cost cut-engineered Marantz player. I mean I prefer the Sony voicing but I'd still take a flagship Marantz over a budget Sony any day.

I think conflating presentation and quality leads to a lot of needless arguments on forums.

That's a great point, I'll keep that in mind. Every day is educational...

[alphaGT]

I had this idea years ago, not exactly the same, but your idea reminded me of it.
So what if you feed the output of a low output moving coil cartridge directly to an A/D converter. Without any modification, just digitize it. Then, add a few zeros to the numbers, to crank up the volume in the digital domain, perform your RIAA in the digital, and then feed it out to a DAC. I thought I may be onto something, but I think that may be how the newest PS Audio phono pre works? I’m not sure exactly what goes on inside the PS Audio box, but I think it’s got an A/D in it. It really is hard to have an original idea these days.

Years ago, before Digital was invented, I thought of a music card. A plastic film, that would have a zig zag path on it that would be dark and light, working exactly like the soundtrack on the edge of a roll of movie film. But instead of moving the plastic card, use a cathode ray to go back and forth, just as it does in a picture tube, and follow the zig zag path on the card, shining through it onto a large photo receptor, and it would play music with no moving parts. But of course digital has removed any need for such a device.

Russell

[Primalsea]

OK, here is some mental gymnastics for you:

Considering that the only purpose of placing a vinyl disc onto a platter that is spinning at 33/4 rotations per minute is to agitate the cantilever whose motions generate electrical signal inside the body of a cartridge, wouldn't it be better if we'd design a sophisticated robotic hand to instigate such motions?

Instead of the blueprint of the recording being stored inside the microgrooves, we could store the blueprint digitally. After all, the vinyl blueprint is merely telling the cantilever when and how to move. Once we agitate the cantilever to move according to the specs, its motion will result in electrical signal which is then transmitted to the phono preamp etc. etc.

The advantage of this hypothetical solution is that we would completely eliminate the relatively brittle signal carrier -- vinyl disc. No wear and tear anymore. And the robotic hand movement will be analog, thus avoiding issues with digital conversion (DAC).

Best of both worlds? But is it doable? Is there an engineering hero who can walk on water and part the red sea and make such an amazing little robot?

If have understood this correctly isn’t this similar to what a speaker does when powered by a class D amplifier? Discrete changes in voltage at a specific time interval cause a transducer to effectively turn a digital signal into an analogue one without a digital conversion taken place per se.

You could also say this is what a DAC does as the output is actually a facsimile of an analogue signal made up of discrete voltages outputted at a specific frequency.

I like the idea, even if it is a bit Steampunk-like compared to what is already done.

I also like the other idea of an algorithm to look at the digital signal before and after the actual point to be converted and to try to learn how to fill in the gaps between samples. (if I have understood correctly). The current solution to do this a brute force increase in bit depth and sampling rate.

Of course we could always just have an analogue signal accurately traced into a groove and get the information off by a mechanical means. Effectively unlimited sample rate and bit depth if a bit truncated.

[walpurgis]

Years ago, before Digital was invented, I thought of a music card. A plastic film, that would have a zig zag path on it that would be dark and light, working exactly like the soundtrack on the edge of a roll of movie film. But instead of moving the plastic card, use a cathode ray to go back and forth, just as it does in a picture tube, and follow the zig zag path on the card, shining through it onto a large photo receptor, and it would play music with no moving parts. But of course digital has removed any need for such a device.

Russell

Didn't Philips (was it Philips?) have a similar idea with laser discs? A stylus tracking the groove but not reading it and laser reflections collecting the signal info. That was non-digital if I recall. This may have been a video system, I don't recollect.

[struth]

Think that was philips analogue optical audio disc..was quite big

[alphaGT]

Didn't Philips (was it Philips?) have a similar idea with laser discs? A stylus tracking the groove but not reading it and laser reflections collecting the signal info. That was non-digital if I recall. This may have been a video system, I don't recollect.

Yes it was I think. I recall that machine, everyone thought it was amazing when it came out.

But my idea has no moving parts. That was the concept. The cathode Ray is the only thing moving, and it is electrically motivated, and an electron beam, not a moving part. It seemed to me back then, that most playback devices were plagued by vibrations, motor noise, tape wobble, wow and flutter, whatever, some moving part was interfering with the accuracy of the playback.

So this was the only way I could dream of to make a playback system with no moving parts out of only analog devices. I’m sure it would have sounded like crap, the audio on the edge of old movies wasn’t all that to write home about. But as with many things, it could have been refined. The electron beam should be able to narrow down the size of the playback point, from the regular lightbulbs that powered the playback in old projectors. And in doing so provide many more shades of gray to the recording scale. I suppose you’d need a similar device to make the recordings? By exposing a sheet of film with an electron beam that would vary in intensity to the music. Develop the film, and you’ve got a tough music storage medium that won’t get scratched up from playing. Fingerprints would probably be the biggest issue. Hey, I was only 15 years old at the time. Or I’d have built one.

Russell

[vintagesteve]

Digitally, according to the o/p.

Which begs some questions...

A lot of questions. Like, why bother? when we already have more than enough perfectly good digital storage systems for musical recordings such as CD, DAT, hard disc etc etc. The whole point of the 'record deck' is that it is purely analogue, and people like that. They like the sound (not sure everyone who claims to can hear a difference, but that's another story!) the equipment, the technology, the 'vibe', the looks and so on.

[magiccarpetride]

A lot of questions. Like, why bother? when we already have more than enough perfectly good digital storage systems for musical recordings such as CD, DAT, hard disc etc etc. The whole point of the 'record deck' is that it is purely analogue, and people like that. They like the sound (not sure everyone who claims to can hear a difference, but that's another story!) the equipment, the technology, the 'vibe', the looks and so on.

Current state of digital audio is based on the deductive approach. Meaning, it is based on the mathematical equation. It is a pure top-down model, where a sampling rate is chosen upfront and then the continuous signal is chopped up in little discrete particles/pieces.

The reason we like this approach is because it simplifies many things (but introduces a few complications in the process as well). However, technology progress never sleeps, and today, 40 years later, we have access to much more sophisticated technology that is becoming mainstream. This is due to the commoditization of the computing power (i.e. the 'cloud').

To cut the long story short, enter Machine Learning (ML). What ML specializes in is the ability to learn inductively, without being given any blueprint upfront. It's a bottom-up approach, the exact opposite from the old school top-down (deductive) approach.

I won't bore you with technical details here, but wanted to point out the ability to let machines observe some phenomenon (such as microscopic movement of a stylus/cantilever combo mounted on a well tempered turntable). After observing that phenomenon, machines could learn how to emulate such phenomenon. Sort of like observing skilled calligrapher do his craft, and learning how to produce similarly beautiful calligraphy.

If you think that is outlandish, not so fast -- three years ago I've worked on a project where we were teaching machines to learn our handwriting, and they learned quickly and were capable of perfectly emulating anyone's handwriting.

[alphaGT]

If you think that is outlandish, not so fast -- three years ago I've worked on a project where we were teaching machines to learn our handwriting, and they learned quickly and were capable of perfectly emulating anyone's handwriting.

The world is changing at the quickest pace ever. In college I missed out on a course on nano robotics, and had to settle for the regular robotics class. back then it was the science of the future. New quantum computers, machine learning, A.I., Nano robots, the next big thing could be truly huge! Imagination is the limit.

Russell

[Macca]

I also like the other idea of an algorithm to look at the digital signal before and after the actual point to be converted and to try to learn how to fill in the gaps between samples. (if I have understood correctly). The current solution to do this a brute force increase in bit depth and sampling rate.

.

Sampling rate determines the highest frequency you can capture and replay. Bit depth determines the maximum dynamic range. Neither has anything to do with the 'gaps between samples'. So called 'hi resolution' digital increases dynamic range and increases the highest frequency that can be captured. So a rate of 96Khz will reproduce up to 48khz frequency. Since no-one can hear over 21Khz, the vast majority of people over 40 will struggle to hear 16KHz, and only vey special studio mics will pick up anything above 22 KHz it is a pointless exercise except for impressing people who think bigger numbers must be better. At that it has been really successful.

Digital systems do not guess at what lies 'between the samples'. In terms of recreating the wave form they are orders of magnitude more accurate in doing so than any analogue source.