Publishing RTTA results?

[Terry McGee]

When Graeme and I released the first system for Real Time analysis of tuning (RTTA), I was concerned that it had great potential to embarrass flute makers and players, especially as even the best tuned flutes can look a bit dodgy compared to accordions and other fixed tuned instruments. So I pleaded for time for makers and players to get used to the new technology, and make changes in their products as they saw fit, before their work was put up for public scrutiny.

But precisely because the results of RTTA need interpretation, others are calling for the "ban" on publication to be lifted. They would argue that we will only learn how to interpret the results if we can discuss them. It's a very fair point.

To put some numbers on it, I had in mind an embargo of about a year - time for makers to embrace the new measurement system and make the necessary number of new flutes to iron out the wrinkles revealed. Graeme, keen to see the benefits of RTTA coursing through the ITM bloodstream, reckons 4 months, which would be up in 10 days time.

You can see how this discussion has developed in the Nach Meyer thread, but I'm opening it wider here to invite in those who have not been following that thread.

The question is, when should we lift the (informal) embargo on publishing examples of flute tuning by modern players and on modern flutes in these pages? Are we ready for it yet? If not, when? Are there rules that should govern us?

Are we generous enough to recognise that we now have a better microscope that the makers of our flutes had when they made them, so it won't be surprising if we see things they couldn't. Are we ready to embark on discussions about what's possible with problem notes like F#, cross-fingered C and C#? Will we remember that an appalling result might be a sign of poor making, poor playing or both?

I can't close this introduction without disclosing that, while I would have enjoyed a year to optimise my own flutes, I agree with Graeme and others that the sooner we get on to discussing all these matters the better. So, count me as a fence sitter.

Let's hear your views.

Terry

[Carey]

Seeing how this data looks for individual flute/player combinations could be very interesting. Also, if you want to make a replica of a particular flute, it should not just look like the original but play like the original - whatever that is, and this tool will help define that.

I don't think the problem will be in exposing the new data that is now available, but rather in the mis-interpretation of that data. Possibly by folks not familiar with looking at telemetry. Telemetry doesn't "tell" you anything. It has to be interpreted. The trick will be in seeing that fair interpretations are made. I don't know what you can do about that. Sooner or later the cat will get out of the bag, and it would be good to have something in place to assist in interpreting the data seen.

I certainly hope that it will not be used to over-simplify complex issues into an arbitrarily binary good/bad classification. It just isn't that simple, no matter how bad we want it to be.

[nitterwhiskers]

When I play a new flute there are many qualities that I assess such as tone, dynamics, responsiveness, tuning, feel, keywork etc etc. When playing to a tuner I can play most flutes in tune. I'm sure my tuning wanders without such an 'aid'. But the fine tuning of flutes is just one aspect of a flutes' playability. Perhaps in the future we can have software that assess tone and responsiveness and loudness and ... And then once we have this 'perfect' flute we can have it cnc machined ...

[groxburgh]

You can download the latest version of Polygraph.r here: http://www.business.otago.ac.nz/infosci ... /basic.zip
I posted it on the web yesterday and it has both chromatic version G3 to C7 , and diatonic D4 to B5 in the one GUI. You can very easily set the defaults because they’re all at the top of the .r script. For Windows you can run it from a batch file (included in basic.zip) or just create a shortcut to it with
"C:\Program Files\R\R-2.6.2\bin\R.exe" --no-save < Polygraph.r
as the target..... Ask if you’re unsure. For the Mac just do whatever it is you need to - I mean Macs are just so easy to use you'll have no trouble figuring it out. And it’s got some other stuff that Terry requested.

As to publishing plots or not: Polygraph/Flutini presents some data. It doesn't mean anything until someone interprets it. And it can easily be misinterpreted but I think it's time to start discussing some of what it's showing.

I don't think I've ever suggested there aren't lot's of other factors besides tuning involved in assessing a flute. But my personal opinion is that it's sad that we even need to be discussing the tuning of modern flutes. It's 2008 not 1828 and we have tools available now that weren't available 180 years ago. Let's use them.

Cheers
Graeme

[dcopley]

This has been a commendable effort and a lot of hard work from Graeme and Terry. I have certainly had a lot of fun with the program, and learned some things about my own flutes and my own playing. My one reservation, and it is a major one, is that we don't have any assessment of the variations introduced by the measurement process itself. In other words, we need to be able to say something like "on this flute, the E note is 20 cents sharp with an error of plus/minus 15 cents". This would be similar to the "margin of error" of an opinion poll.

I recall there was some earlier discussion of this topic, and there were some proposals to get at this by analysing a number of players on a systematic basis. I did a mini-study myself, by playing and recording a tune six times through, then processing the results. My most variable note was the low D, which ranged from 13 cents flat to 9 cents sharp. The least variable notes had a range of 7 cents over the six trials. The overall standard deviation averaged over all notes was 5 cents. I would expect that if you expanded the study to more players and instruments, the variations would be larger (unless I am a particularly inconsistent player!).

My suggestion is to hold off on publishing results and conclusions for particular flutes and makers until this question is answered. Until then, we really don't know how much confidence to place in the measurements.

Dave Copley
Loveland Ohio

[groxburgh]

My suggestion is to hold off on publishing results and conclusions for particular flutes and makers until this question is answered. Until then, we really don't know how much confidence to place in the measurements.

Dave makes a very good point about not knowing how much different players (or even one player) will vary on a given instrument. But I see that as a good reason for publishing now not as a reason for holding off until we know more about it. Let's share what we do know not hide it because there's still stuff we don't know. Sharing plots will enable us to determine how much confidence we can place in the measurments.

We can not make the claim that a flute is bad just because 1 plot or even several show really bad tuning - and that's good for makers. But what we can do is look at a plot of someone playing a particular flute and say if you alter how you play (eg, cover more of blow hole and blow down more) or if you alter your flute (eg wax in L2 hole to lower B) you'll be much more in tune and enjoy playing with others a lot more.
Cheers
Graeme

[Terry McGee]

Dave raises interesting questions about repeatability. First thing I guess is we need to separate the measurement system and the player. Graeme, have you (or could you) run the same recording several times to confirm you get the same results (within reasonable bounds) every time when you feed in exactly the same sounds?

My guess (without knowing about the RTTA systems' repeatability) is that repeatability is a player issue. I know I get different results if I play at full tilt rather than demurely. So that is something that we have to learn to live with. It doesn't diminish the usefulness of RTTA, indeed it confirms how important it is. Without it, we could be tuning for the least likely set of circumstances.

I think there will be a number of matters that we will have to come to understanding on if we are going to reap the benefits RTTA offers, rather than getting sunk into confusion. But they will require the input of many, which I think is an argument in favour of getting on with public discussion. Gulp.

I'm hoping that we can all agree to be mutually supportive! If maker x finds that most people seem to play his/her bottom notes flat, or top octave sharp, or whatever, that shouldn't be a cause for derision. Presumably this will be new knowledge to the maker as well as probably to the players (or customers would have reported it to the maker earlier, and the maker would have already made changes).

I think indifference would be what should attract derision!

Terry

[groxburgh]

Graeme, have you (or could you) run the same recording several times to confirm you get the same results (within reasonable bounds) every time when you feed in exactly the same sounds?

Yes Terry it gives the same results every time. You know this. I've just fed in Matt Molloy playing the Mason's Apron and it gives a plot like you'd expect - in tune, (but at about A=448) I've just plotted Hammy playing a dozen tunes and it shows him playing in tune. Now if it can cope with the best ITM players playing at ludicrous speed I think it's time to just start believing what it shows.

Who or what are we trying to protect by not publishing plots? If it's the makers I want to know why?

Cheers
Graeme

[Sillydill]

I'll raise a Red Flag on FLUTINI !

I've encountered problems on some flutes that have rather strong partials. FLUTINI registers the partials and sometimes not the fundamental note.

TARTINI does not appear to suffer this problem.

[groxburgh]

Well actually Tartini does suffer from this problem as well as Flutini. There is a setting which affects it but sometimes it happens anyway. It doesn't make the results invalid but does emphasize the need for caution when interpreting plots.
Cheers
Graeme

[Terry McGee]

Graeme, have you (or could you) run the same recording several times to confirm you get the same results (within reasonable bounds) every time when you feed in exactly the same sounds?

Yes Terry it gives the same results every time. You know this. I've just fed in Matt Molloy playing the Mason's Apron and it gives a plot like you'd expect - in tune, (but at about A=448) I've just plotted Hammy playing a dozen tunes and it shows him playing in tune. Now if it can cope with the best ITM players playing at ludicrous speed I think it's time to just start believing what it shows.

Heh heh, just checking. So we can be confident that differing results when we play is a statement of how we are playing, not how the measurement system is doing its job. So then it becomes a matter for the player to come to grips with.

Who or what are we trying to protect by not publishing plots? If it's the makers I want to know why?

Cheers
Graeme

Makers certainly. Their livelyhood depends on what people think of their instruments. It's legitimate for consumers to discuss publically what they think of various makers providing they are accurate in their comments. Because players are going to have to learn how to interpret RTTA results we might expect some knee-jerk reactions - "Hey, my new McCrud is a whole cent flat on F# - I'm sending it back". We're going to have to be ready to hose some concerns down. It will be at times a complex discussion, but an enlightening one, so we shouldn't shirk it.

There might be some uncomfortable moments for professional players too, when they first learn by reading Chiff and Fipple that they are not coping as well as they thought with the inherent tuning of their flutes. That seems less likely to impact financially on a player than a maker - I don't see major concert promoters turning to RTTA to decide on whether or not to book Matt.

I'm looking forward to a lot of this discussion. For example, I'd be interested in exploring with others how we feel we should best handle the sharp c, flat c# compromise - there are some very interesting issues to consider - I could see a whole thread on that one alone.

Leaning further over to Graeme's side of the fence ....

Terry

[groxburgh]

Who or what are we trying to protect by not publishing plots? If it's the makers I want to know why?

Cheers
Graeme

Makers certainly. Their livelyhood depends on what people think of their instruments. It's legitimate for consumers to discuss publically what they think of various makers providing they are accurate in their comments. Because players are going to have to learn how to interpret RTTA results we might expect some knee-jerk reactions - "Hey, my new McCrud is a whole cent flat on F# - I'm sending it back". We're going to have to be ready to hose some concerns down. It will be at times a complex discussion, but an enlightening one, so we shouldn't shirk it.

Terry

All a maker needs to do is publish a plot of someone playing the questionable flute in tune!
Cheers
Graeme

[Terry McGee]

Recorded and analysed before witnesses!

Terry

[Carey]

All a maker needs to do is publish a plot of someone playing the questionable flute in tune!

Easy for you to say.

Seriously, you know that, and the rest of us here know that now, but if you are a maker and religously check your e-mail once a month whether you need to or not, how will you feel when you find you've gone from hero to goat in the space of a week or two? And all you have to do is install some new fangled software on your Windows 3.x machine and figure out how to use it. Get a player to play a flute you no longer have access to. All the while having customers who expect their flutes to be delivered.

OK, that's a bit over the top, but I think that's where the down-side is. Not that it changes the decision, but I think that's part of the reason to be careful, to have some damage control strategy before giving Pandora the key.

Perhaps the wait isn't a wait, but a period of time when the pioneers get out in front, take some arrows in the chest, and get the market generally comfortable with the plots before it's turned lose. Nobody is saying NO flutes will be named. Only that no makers should be named without their permission. I for one am already using it on my whistles, and it might cause me to release another model or two in addition to the current models. We'll see. I have to play with it a bit more. But I'm already enlightened. (Not that I'm any brighter, if you follow what I mean.)

Carey

[monkey587]

What flute was Molloy playing on that recording? Was it one known to be in "in tune" by your definition? Wouldn't his trusty Boosey pratten have a pretty flat C#? What about Hammy?

A month or two back I got annoyed at the tuning of the high A (20ish cents sharp) on my flute, and used Graeme's wax trick to bring it down some. It worked, and the A was easier to play in tune. I sent Graeme a PM and thanked him for that. I have since undone it.

You see, I was frustrated that the C# was flat (40ish cents) on my other flute, which also has a sharp A. In fact, before tweaking the one flute, its tuning was basically identical to this second flute. In response to a comment he'd posted elsewhere, I PM'd Brad Hurley and asked for some advice on it. This led me to spend a some time each time I practiced playing different combinations of the high G A and B as well as combos involving the C#, making sure they were really well in tune to my ear. After a few days of this, playing these notes in tune and at speed was significantly easier, my embouchure was much more sensitive, and frankly my tone was a hell of a lot better. Am I perfectly in tune all the time? No, but I'm a lot better and will continue to improve, and I was able to better match my sessionmates including those with fretted instruments.

I believe that the implication that flute makers no longer have any excuse not to make "in tune" instruments, I think, is pretty lame. Terry is hesitent about publishing results, but as good-naturedly as possible and probably unwittingly has already begun casting a shadow of pity/doubt over all those makers who were unfortunate enough to make instruments without the benefit of RTTA.

If you want to make perfectly equal temperament flutes, by all means go ahead. I predict that they will tend to attract less experienced players with undeveloped embouchures, which will likely remain undeveloped. They will show up at sessions in the hands of newbies who will have poor tone and likely sound quite out of tune. More advanced players will probably find that they lack color and won't bother with them. People will groan when they turn up at sessions, just like they do with the high end whistles, whether the owners of said whistles like to believe it or not.

There are SO many players who play old rudalls and prattens, german flutes, and modern "out of tune" flutes, who manage to play well in tune and sound great doing so. And they never seem to complain about their instruments, either.