Piano World Home Page Forums Our Most Popular Forums Piano Tuner-Technicians Forum Are the Days of Human Piano-Tuners Numbered?

Iknow this has been discussed many times before, just found it interesting to see it on WSJ...

From the Wall Street Journal Blog

http://blogs.wsj.com/ideas-market/2...no-tuners-numbered/?mod=google_news_blog


By Christopher Shea

Getty ImagesIf you just wanted to play in one musical key — say, C — then an electronic device could tune a piano better than any human. But since the widespread adoption of “equal temperament,” a process through which every musical key is rendered ever-so-slightly out of tune in order that every key can be more-or-less in tune, piano tuning has been a craft — a complex process of managed imperfection. That’s why, every so often, piano owners have to pay someone a couple of hundred dollars to tune their instrument.

But now a German physicist has developed an algorithm that allows for the automatization of equal-temperament tuning. Where a human tuner relies on her ear, experience, and feel, the new process views tuning as “problem of entropy minimization.”

“That may cause some sleepless nights in the music world,” observes the Physics arXiv blog. “It may well be that the professional tuner’s days are numbered.”

Thanks for the link, but BS. Typical hyperbole. No one is going to lose any sleep, let alone have less work because of someone at an office crunching a few numbers and coming up with a software application.

It doesn't look like this is much different than standard ETDs used widely today. An improvement? Perhaps, that would need to be seen in practical and commercial use. But for certain this is nothing ground breaking as it is described.

ETDs cannot tune a piano anyhow. Turning the pin properly is just as important, maybe more, than identifying the correct pitch of a given note.

I hate to be the bearer of the obvious…but tuning algorithms of one sort or another that take inharmonicity into account have been around for some time. They are contained in what are commonly called “Electronic Tuning Devices” (ETDs) and are currently in use by many piano technicians. As at least one of these (TuneLab) is available as a free download I suspect a few owner/amateur tuners are using them as well.

The difficulty is not so difficult coming up with a pitch standard, the difficulty—as will be attested by any tuner still working on his first one hundred or so tunings—is in getting the string to that standard and keeping it there.

ddf

Golly gee. Add this to the piano that tunes itself and I guess we're just doomed.

An interesting article, but typical of Journalists-Reporting-On-Science.

There is a lot of missing information and huge mental leaps that are happening.

First of all- just on the scientific/engineering front, there doesn't appear to be any "machine" made at all. Just the development of an algorithm -- which is just an impressive term for "best shot guess". Some of those "best shots" are pretty good. Most modern ETDs (except the strobe technology) rely on an algorithm. So, it's just software -- actually not even software but a claim to software. And we all know about about software that will be available Real Soon Now . . . its called "vaporware".

But beyond that, there's the question of whether there will be a "machine" of some sort or not. Whatever-it-is will have to be operated by a person - just like current ETDs. Conceivably a factory might invest in a robotic process, but that won't happen for pianos in other locations. Can you imagine dragging such a machine around from piano to piano?

And then there's cost. I think most anything can be built. But at what final cost? The total number of piano technicians in the world is not a good base for economies of scale to cut in. Maybe it could be built from off-the-shelf parts??

So, bottom line, nothing exists yet except an idea or at most a software routine. That's pretty close to nothing at all. Then you have the steps to reality which include prototyping, production, marketing and adoption. And nothing whatsoever to indicate that the use of such an item would provide a lower price to consumers for an actual piano tuning service call.

I don't think any professional technician reading this -- nor even someone contemplating entering the field -- needs to hold their breath on this one.

Articles written by folks who don't know the piano tuning field are best regarded as entertainment, or light reading for the general public.

As Del and Supply said, That wire has got to be moved by a human and moved so it will stay put for a while.

I will add that it is a sure sign of poor journalism to not do some background research in the field to which the topic relates.

Very often, in small newspapers etc it is fairly easy for anyone to send out a press release and have it printed as a news item. Perhaps this trend is taking over into more specialized press as well.

A simple bit of research would have brought up what we all know - that there are a lot of very well rolling wheels out there, and that now someone has managed to re-invent it yet again. Hardly revolutionary.

I would not mind an apprentice right now, young and willing.

Plenty of work, a career as an entrepreneur. There are plenty of pianos that still have life in them, many end up in a landfill, a shame.

The piano business has grown smaller in the last few decades, so less piano technicians have been needed, and more young people mistakenly dismissed it as a dying trade.

The piano, as we know it today, will always exist, same as a cello, a fiddle etc... It has become a classic. Tuners will always be needed now.

I hope new acoustic instruments will come along, but that is another story.

I am sure the person who wrote the blog post never bothered to try to tune a piano with the equipment he was hyping.

That's Funny! Sometimes when people see my tuning program they are fascinated and then sometimes I'll say, "Hey this is the 21st century. I'm just lucky that a robot isn't tuning your piano!" But all seriousness aside... What kind of robot or machine would it take to manipulate all the pins? And then hooking it up to the piano. It just seems really impractical.

I'm not clear on just how all of you guys feel about this...

It's okay, you don't have to gild the lily...

I, for one, think this is simply interesting. Any investigation into string vibration can deepen our knowledge of what's going on.

I don't think we are threatened more as piano tuners, than any other profession which relies on technology for its subsistence. As long as there are traditional instruments out there, there will always be the need for technicians. If the instruments evolve, there will only be a new need for technical expertise.

For those with a more academic interest into what Hinrichsen was researching, rather than the reporter's misinterpretation, here is the Cornell University archive of the paper: http://arxiv.org/abs/1203.5101

He states some important things, which I hoped the piano community would have seen. Of them:

- ETDs use stretch algorithms based on statistics of best fit tunings

- ETDs use bandpass filters for single partial alignment to apply these algorithms

Both are true reasons for the shortfall of electronic tuning.

By minimizing entropy over the whole tone, he is indeed getting closer to what good aural tuners do.

He also makes the insight that excellent aural ET tuning over the Railsback curve, results in unequal intervals, which are not random. He believes they are patterned to the specific para-harmonicity functions of the instrument.

Here is a figure from pg. 10 of the paper (basically, the enlightening outcome of the research):



Remarkable to me is how, for the first time in any published scientific literature, I am seeing something remarkably close to what aural tuners do. This 'unequal' temperament has been scoffed at, as some kind of human error that ETDs have subsequently improved upon.

We can see here in clear scientific demonstration how this work has produced very similar unequal deviation that incredibly good aural tuning produces: the red curve generated by his entropy reduction Monte-Carlo scheme and the black, by a skilled aural tuner.

For those who don't want to read through the entire paper, here are some important points:

"Measuring the frequencies of an aurally well-tuned piano one finds that the tuning curve is not smooth, it rather exhibits irregular fluctuations from note to note on top of the overall stretch (see Fig. 1). At first glance one might expect that these fluctuations are randomly distributed and caused by the natural inaccuracy of human hearing. However, as we will argue in the present paper, these fluctuations are probably not totally random, instead they might reflect to some extent the individual irregularities in the overtone spectra of the respective instrument and thus could play an essential role in high-quality tuning. Apparently our ear can find a better compromise in the highly complex space of spectral lines than most electronic tuning devices can do.

Fig. 8 shows the resulting tuning curve of a typical run compared with the actual curve produced by an aural tuner for an upright piano. As can be seen, not only the overall stretch is predicted correctly but even the fluctuations of the two curves are highly correlated, especially in the bass and the treble. Apparently the entropy-based tuning method is capable of generating the same individual deviations from the average stretch as an aural tuner. This is surprising and not yet understood, but it indicates that these fluctuations are reproducible and may play an essential role in the practice of tuning.

The implementation of the method is very easy. The tones are recorded, Fourier-transformed, mapped pointwise by the psychoacoustic filtering functions as described above, binned logarithmically, added up, and finally plugged into the entropy functional.

An explicit identification of higher partials and the measurement of the inharmonicity is not needed. The method is expected to take automatically any anomalous spectral properties of the instrument into account.

However, the method suggested here is still in an immature state. It could be modified in various respects and a systematic study is still outstanding."

If Hinrichsen's methods were implemented in ETDs, it wouldn't do anything of the sort of putting tuners out of business. It would simply make the ETDs a little better.

Tunewerk,

Thanks for posting the key points of the actual paper. Very interesting! (And quite impressive.)

When I was learning to tune aurally, I apprenticed with Steve Fairchild, RPT here on Long Island for several years. Steve was a prominent aural tuner who actually got into the Guinness Book of World Records for piano tuning, and worked with Sanderson on aspects of the SAT (He created the magnetic pickup used with the Accutuner). Steve had an intimate knowledge and understanding of the physics behind the tuning process, and was adamant about conveying that in his teaching. I remember that he gave me a series of computer disks containing a program he had developed over many years that actually took the FAC approach many steps further by sampling 37 different notes throughout the scale to determine inharmonicity and create a custom stretch. This was back in the 90's, when the Accutuner was the only game in town for ETD. The interesting thing to me is that a man like him, who was obviously a top notch aural tuner by any standard, would spend years working on a computer program to tune pianos. From my conversations with him on this, it was clear that he believed that a well conceived program could exceed the performance of even the best aural tuners. Of course, you still require a craftsman to translate the results of the computer analysis to a viable, functional, stable tuning.

Even if there were a machine or robot that could tune your piano, how would it get there? Take the bus?

This addresses a question I was thinking about. Why did these people even invent these devices? Did they not foresee that there would be tuners that would rely heavily on these devices or programs? Were they just trying create a tool that would be helpful in tuning but not to be relied upon? If it's there, somebody is going to use it. Seems to me if they had really wanted to promote aural only tuning they would not have invented and promoted these things. So I think the only question that remains is; To what extent do we use these ETD's? Are they only for pitch raises and rough tunings or can they be competently used or relied on to tune a whole piano?

My answer to this would be yes, absolutely. My whole purpose of bringing this up is because these programs were created by already great aural tuners, who had no reason to spend the time and effort to do this unless they thought it could be an improvement over what was already available. At least that was the case with Steve, from my discussions with him.

With the greatest respect to both aural and ETD tuners, it doesn't matter which is being used just as long as the tuner has the REAL SKILL of setting the pin. There are good, bad , and plain ugly tuners in both camps ... what was it that someone once said ? those that can .... do .... those that can't .... teach ! The very best examples of both aural and ETD tuners still require vast amounts of experience and skill to be able to tune to the highest standards with rock solid stability. Those without the experience and skill can only hope to achieve a high standard over time. A well trained ear or top of the range ETD will never be able to compete with a human brain that has mastered the finesse of pin setting.

It's quite easy for anyone with a good ear or ETD to experiment on tuning a piano, but quite a different matter to obtain a tuning that stays where it's put

I understand your point that they thought it might make an improvement. ETD makers are on a search or pathway and have made great strides. But they are not "there" yet. It may be that they - as is the case in other fields, as well, were dazzled by the promise of technology and underestimated their own prowess as aural tuners.

Evidence for this idea is simply in the continuing introduction of new models (same brand) of ETD. I've gone several times now around this merry go round:

1) tuners show off their "sparky-matic whiz-bang ETD" that is at the apex of technical art.

2) I follow up after users of said latest device mentioned above and, after a few times, recognize deviations that don't match what I'm hearing.

3) tuner then gets the "Mark II" version of the sparky-matic and again proclaims it to be the apex of tuning capability.

4) In follow-after tunings, I note improvements but still things that are off.

5) Rinse and repeat.

My stance on ETDs:
1) I love tech stuff so I'm not at all anti tech.
2) I know folk that use them and they produce excellent tunings
3) As long as I can hear inconsistencies, I cannot convince myself that they offer me anything of benefit.
4) I understand what "algorithm" means -- which is somebody's "best guess" at the relevant variables and their interaction for any given tuning scenario. Algorithms have become very good, but sometimes "best guess" doesn't equal on-the-ground reality.
5) I don't even think that the variables that affect good tuning have even been fully identified. Capleton's discussion of "soundscape" is an excellent treatment of that topic.

Johnkie,
With all due respect, while I quite agree that it takes a vast amount of experience and technique to "set" a great tuning, this whole concept of "obtaining a tuning that stays where it's put" has always seemed rather nebulous and arbitrary to me. For starters, what does the term "stays where it's put" mean, exactly, in terms of time? If I set a tuning that lasted only an hour after a player, of any level, played the piano, could that still qualify as a great tuning? How far out would it have to go to then be considered less than "top notch"? For instance, I see that your signature line indicates that you are a concert tuner. At every concert that I've ever attended, the technician has always come out at intermission to retune the piano. So, doesn't that mean that, essentially, his/her tuning didn't hold? Does that, then, make him/her an inadequate tuner? Oh, I know that the response will be that the professional level pianist is playing the piano harder, and is more demanding, than the average pianist, so that then brings us back to the question of what, EXACTLY, is the metric by which we measure a "great tuning" from the standpoint of stability? The other consideration is the fact that the tuning pin is just one component that contributes to a piano going out of tune. The others, of which I am sure we are all aware, have little to do with what we do with the hammer on the pin. If the pinblock is worn, if the humidity is allowed to swing excessively, if the strings are new, if the player is a "banger", if the piano is being used 8 hours per day, all these factors are out of our control.