Piano World Home Page Forums Our Most Popular Forums Piano Tuner-Technicians Forum Entropy Piano Tuner

Deterministic Newtonian mechanics do not apply to the white noise produced by a simple diode (shot noise). I have been building and using these photonic quantum devices for over 50 years. They are truly random.

Edit: Sorry Isaac!

How do you know they are "truly random"? Just because
current science cannot measure an electron's position
and velocity at the same time? What if the noise only
appears to be "random"?

From http://physics.info/motion/:


"Some motion is predictable in theory but unpredictable in practice, which makes it appear random. For example, a single molecule in a gas will move freely until it strikes another molecule or one of the walls containing it. The direction the molecule travels after a collision like this is completely predictable according to current theories of classical mechanics. Every measurement has uncertainty associated with it. Every calculation made using the results of a measurement will carry that uncertainty along. Now imagine that you are trying to predict the motion of a billion gas atoms in a container. (That's a small amount, by the way.) After measuring the position and velocity of each one as accurately as possible, you enter the data into a monstrous computer and let it do the calculations for you. Since the measurements associated with each molecule are a little off, the first round of computation will be a little wrong. Those wrong numbers will then be used in the next round of calculation and the results will be a little more wrong. After a billion calculations, the compounded error would render the results useless. The molecule could be anywhere within the container. This type of randomness is called chaos.

Some motion is unpredictable in theory and is truly random. For example, the motion of the electron in an atom is fundamentally unpredictable because of a weird conspiracy of nature described by quantum mechanics. The harder you try to locate the electron, the less you know about its velocity. The harder you try to measure its velocity, the less you know about its location. This is fundamental quality of small objects like electrons and there is no way around it. Although the electron is often said to "orbit" the nucleus of an atom, strictly speaking, this isn't true. The probability of finding the electron at any particular point in space is predictable, but how it got from the first place you observed it to the second is actually a meaningless question. There is no name for this kind of motion because the concept of motion doesn't even apply."

I would argue with the statement that this is "truly random".

More like: We don't understand this yet, or lack the tools to measure it properly.

God does not play dice, but even if He did, the outcome can be predicted!

**Crackpot alert**

Kees

Stay on topic and cut out the personal insults.

what 'personal insults ' ? I see no personal insult at all

While the results are different each time the calculation is performed, they seem, in the trails I have conducted, to converge on the same basic tuning, the differences being likely inaudible.

yes I have read that

you could compare with excel now, I do not have enough time to do so.

each new computation allow to create a csv file.

I tested the midi feature, but due to the poor sound card of the tablet pc, the results are not very easy to check.

I heard fast beats in twelves, which is unusual, but thos is inconclusive at that point.

Finally downloaded on an older netbook. It's really slow to calculate! Sampling went all right. Didn't try to tune yet, just getting more comfortable with the software. Is there a way to see the numerical value for the calculated offset along with the targeted partial number?

As with all tuning software, the tough test for me is to use it to tune a smaller piano to see how it deals with octaves across the break and into the bass... Should be interesting!

Ron Koval

Hi Prout,

Yes, I understand what you mean, though the question would then be: "inaudible" who for?

Perhaps you would answer for your ear, or for a vast number of ears, but it is the idea that small differences can make a significant difference, also in this case...that is the reason for my investigations.

The same could be said, "likely inaudible" differences, when comparing this EPT with other ETD's, though (IMO) this kind of conclusion is not convenient for pro tunings.

There must be some kind of variable that makes results different for every calculation; I would like to know what that variable is, and I do not think it depends on the random processing of data. Can it depend on the relative number of calculations? Converging more and more to a "minimum entropy" order?

Does partials "amplitude" play a role?

What is your idea?

Best regards,

Alfredo
.

For a tuner, the varying results would be audible. For listening to music, probably much less so.

Partial amplitudes play a big role in the calculations. A very weak 4th partial and a strong 6th partial on a given note would have an effect on the calculated result.

Also, the recording ideally should be conducted using a perfectly flat response microphone over the desired frequency range. These results are then amplitude adjusted (basically put through a filter mimicking the normal human hearing frequency response - probably an 'A-weighted curve') and the output fed into the interactive calculation algorithm.

Will give this a try this week on my M&H BB.

I wonder what this software lacks in order for the end result to be called a
"professional tuning"? Hopefully will find out when I Iune my BB.

To Ron (sorry I missed 'quote ') yes download the last version if you did not.
Lot of data available in csv file, easy to export to a spreadsheet.

In the main menu, sub "export"

What is surprising is that I think your final tuning is recorded.

Or at least the ability of the software to synthétise "piano tones' using the recorded, or the computed data makes it possible.
I thought that even the final tuning (red dots) can be synthesized and heard, but I am may be wrong on that.

Some very nice ET sounds coming from my M&H BB with Entropy....when I get the tuning stable, will record. So far, I am impressed!

Looking forward to your experience.

I went ahead and tried it on a Chickering console - you might know the kind with both wound and unwound doubles on the long bridge? These kinds of pianos are my standard challenge for any new approach or software to see how it really handles difficult choices...

Ran into beating octaves, so I left off for now... The temperament and going up was almost an exact match for how my Verituner approaches the puzzle, so it is a good start for a software approach.

Ron Koval

Hi Ron,

Ok... I decided to try my outboard Samson USB mic instead of the built in mic of the Laptop....not sure what the difference might be, but on their video and the manual, it suggests using a better quality mic.

http://www.samsontech.com/samson/products/microphones/usb-microphones/gtrack/

So far, I am getting a very nice sounding piano, very resonant and sounding like a good ET tuning. Octaves seem clean at this point.

Btw, I just noticed that it's now available for the USA Iphone/Ipad...I just dloaded it to my Ipad 2...but will stick with the laptop for now.

I tuned my Nordiska 120 today with Entropy.

I am not a professional, but here are my observations:

1) Smooth and even ET.

2) Octaves (and the compromises involved) are good, better than I've been able to do with TuneLab.

3) At first I found it hard to tune accurately, no spinning wheel and no TuneLab phase display etc. But eventually I got the hang of it, and aiming for the horizontal line in the middle of the display proved accurately enough.

4) I miss my UT. For the last year or so, I've had Koval EqWell on my piano, and even though it is a mild UT, it is still noticable that I now have ET.

I know that aiming for low entropy will naturally lead to ET, but I still think it should be possible to impose the boundary conditions of an unequal temperament unto this mathematical problem. In general, it would be nice to have more tweaking options. I know that the basic algorithm only has one optimal solution, but it would still be nice to be able to impose different kinds of boundary conditions such as UT, partial matching etc. as long as you know you stray from the "optimal" solution.

Yeah, the display reaction time was different for me too. Reminded me of OnlyPure, how it seemed to tune deep into the note, rather than quickly. Might be something on purpose, a more European approach?

Ron Koval

Hi pinkfloydhomer,

For several (perhaps obvious) reasons, your so nicely shared observations cannot tell how close to my favorite ET the EPT tuning would sound to my ears.

Would it be possible for you to post a recording of chromatic thirds, fifths, octaves, 10ths, 12th, 15ths and 17ths?

Alternatively, could you post the screen graphs, like Prout did some time ago, showing the previous and the actual tuning curve?

Cheers,

Alfredo
.

That's disappointing that you had bad results across the transition to wound strings.

GPM had great results apparently on his high quality instrument. It seems that the choice of tuning software is "almost" irrelevant when it comes to tuning quality pianos. The real test is on the spinets and consoles.

The ultimate test would be to see which software gives the best out-of-the-box tuning on a Melodigrand spinet

I can't promise that I'll get around to posting recordings of progressive intervals. But let me emphasize this:

The piano I tuned was a cheap 120 cm Nordiska (Dongbei) upright from around 2004. It might not be as bad as the Chickering console or the Melodigrand spinet, but it is far from being a high quality instrument. And I've never been able to get a satisfactory compromise of octaves in the bass/tenor and across the tenor break, or a stretch in the bass that I found musical. I've been using TuneLab and have tried various partial pairs and tweaks, but I always came up short. With Entropy my Nordiska 120 didn't suddenly turn into a Steinway D, but it is the best bass stretch and bass octave compromise I've heard so far on this piano. Even when a professional tuner tuned it with Verituner, the bass wasn't as nice.

The one thing I _really_ miss, is UT.

I intend to make a custom tuning curve for TuneLab that matches this stretch as well as possible. Then I can maybe keep the stretch (or close to it) and still interject UT into the equation.

The overall algorithm of Entropy seems to work well, I just would like all sorts of tweaking options. UT, less weight being put on the extreme ends of the piano (I mostly play Bach, that is C2-C6), going for low entropy but with special emphasis on pure 12ths etc etc etc.

Also the software as such seemed a little rough. I used the Android version, and it didn't always behave as a natural Android app. The most annoying thing was to have to recalculate everytime I reopened the app or loaded my tuning file. Why not save the result of the calculation?

Also, I really feel that some piano tuner developer should make a serious effort on unison tuning. I know that with most notes, it is faster and better to do it by ear. But with some unisons, you have to make a compromise. False beats, indistinct pitch, coupling effects etc. It would be nice to see a "minimize entropy" algorithm working only on three string unisons. Low entropy when you get as clean a unison as possible that is still on target. I would mostly use this feature in the treble or high treble on my piano. TuneLab is precise enough with it's phase display to let me tune problematic unison strings individually and then end up with as nice a combined unison as I am able to. Entropy is not accurate enough for this. You can try to aim for the horizontal line with all three strings individually, but the combined unison is not always as clean as possible.