Piano World Home Page Forums Our Most Popular Forums Piano Tuner-Technicians Forum Setting an aural stretch to an ETD

Hi everyone, I´m a learning about tuning and already have my own tools and been practicing on some pianos. I also have Tunelab which gave me a better understanding of what I was doing aurally and giving me feedback on my hammer technique. I should also add that English is not my first language.

So my idea came from the fact that ETDs
1) Are good for objective decisions such as dividing an octave into, for example, twelve equal half steps for tuning ET.
2) Are bad for making a compromise, like the one needed for tuning octaves which have a wider range of acceptability.

So if you begin with A4 440Hz and then tune all the octaves down to A0 and up to A7 you should have a fairly good idea of the stretch for that particular piano, and of course you have checked that aurally so you use your own criteria for a "good" sound, which is where ETDs fail, especially with small pianos.
Then, if the ETD (Tunelab) has the option of editing the stretch curve you should make curve fit the offset of all A´s, so the program will calculate a very accurate division for all octaves.

I haven´t had the chance to actually try this, but I would like to hear some thoughts. I also know that iH is not constant even within one single octave, but using the offset of every octave should at least avoid dragging error through the whole piano.

Probably the ETD could calculate a better tuning curve, but in the case of spinets where is more difficult, could this be a valid approach? Also, could doing the same procedure with two notes (say, A´s and E´s) instead of only one help reduce the error within the octave?

Thanks

Matias

It is possible to measure and put into the Tunelab file memory the inharmonicity of every note on the piano.

I believe this will give you the most accurate stretch that Tunelab can offer.

Tunelab isn't set up to use that additional data to change the calculation in the way you might hope... It still forces a smooth(ish) curve onto a single partial(of the technician's choice) based on the projected interval match for each section (also of the technician's choice)

The short answer is yes, that is a good plan. The longer answer is that at the difficult part of a challenging scale, Tunelab might struggle to make pleasing 'ear' octaves... Usually it is the lower plainwire going into the wound strings. If you set your octaves, avoiding that section to have a harmonious blend from bottom to top, then you can have a stab at fudging the placement of that half octave or so to fit with what really works above and below.

Set your samples, lock them in using Tunelab and those notes will highlight in red on the graph. Then use the arrows to alter the graph(and the calculation) to get as close as you can to the red samples. Tune away!

Best of luck

Ron Koval

The problem I have with tuning all the "A" notes by ear is that I can't do that without using other notes as a reference. In other words, when tuning by ear, the stretch develops as I tune the piano. Stretch is not pre conceived. Even room acoustics play a part. As I tune up the piano and compare my ear to Tunelab, there can be areas that match, and areas that don't, especially in poorly scaled pianos, or those with awful voicing.

There's another easy way to test the stretch you're using, without being an aural tuner. Assuming you're tuning from bass to treble, stop every once in a while and play the notes one or two octaves below.

In other words, when you get up to A5, play A4 and look at the result. You're now measuring the second partial of A4, and this is the one that you want to match to A5.

As you go higher up the keyboard, the upper partials effectively disappear, so just play the note two or even three octaves lower.

--Cy--

Thank you all for your answers. They have got me into new ways of thinking this issue.

Prout: I know I could measure iH for every single note (in fact i´ve already done so), but as we know, ETDs lacks criteria for some things which I thought might be improved with some aural indications, thus making a "mix" of both approaches (like Baldassin suggests in his book On Pitch).

Ron: If I understood what you said, the most problematic part would be "transition areas", specially on the change from plain to wound strings. Yes, that would require an extra aural effort I believe.

Bob: Since I´ve never tried this strange way of beginning a tuning you made me realize it would a bit awkward to do a compromise such as tuning octaves without any other note to compare. But that doesn´t mean this method would be obsolete, only less accurate than what I initially thought.
Another thought: If you are capable of doing such good octaves out of nowhere then you should also be able to tune all octaves aurally after setting a temperament. But I think this is an interesting discussion at least as an experimental theory.

Cy: What you are suggesting is a slightly different method than what I propose (begin with electronic tuning and then check it, instead of beginning with aurally tuned octaves) although is a totally valid advice as well.

I should add one question related to Tunelab´s use: You can adjust total stretch (on A0 and C8) and also the curve distribution. But are those only two parameters enough for fitting a specific offset? What I mean is that there is not much possibility of manipulation, or is there?

Tunelab is only as good as the information you give it. It is my opinion that the most recent version of Tunelab does an excellent job of establishing a good tuning curve, even on poorly scaled pianos. Here are a few of the things I do to ensure that I'm giving Tunelab the best info possible.

1. If the piano is more than 20 cents off, I don't take iH readings. I simply use the "average" tuning file and do the first pass with over-pull. I then take samples when the piano is closer to pitch.

2. I sample all of the C's & F's up through the 6th octave. I then sample the notes on each side of the break. If the difference in iH at the break is high enough, Tunelab will then turn on the split-scale mode. In split-scale 6:3 octaves in the bass are "highly recommended." I take a look at the iH constant values to see if the numbers make sense. Generally higher numbers at the ends and smaller numbers toward the middle of the piano.

3. I set the auto-partial selection to encompass A0 up to the highest wound string. With this feature, Tunelab will select the strongest partial for each bass note and use that one to tune.

3. After Tunelab has established the new tuning cure, I then toggle through the different octave types to see which ones behave the best.

4. I will then layer the offsets for either the EVBT3 or Ron Koval's Equal well-Temperament on top of the tuning curve. (I prefer a mild well-temperament to a strict ET)


Using these steps, I personally have not found it necessary to aurally tweak the tuning curve.

Assuming you're using the latest version of TuneLab, you have two sets of controls: (1) The gross level of the bass and treble octave styles (e.g. 6:3 and 4:1), and (2) the manual editing of the tuning curve, which lets you make fine adjustments to the curve in four different segments. These are pretty powerful together: I've constructed tuning curves to match two concert grands with a 6' grand performing together.

Read the documentation about the deviation curve at the bottom. Until you get more comfortable tuning by ear, just adjust the two octave styles to get it as flat as you can. In other words, if your choice makes a curve swing in the middle by more than about five cents, try another octave style. Smaller pianos generally work better with narrower octaves (6:3 in the bass instead of 8:4; 2:1 treble instead of 4:2).

--Cy--

I should study more deeply the different octave styles for the tuning curve and also the deviation curve. I´ve read the documentation on that and still can´t get it quite right. But I think it has to do more with understanding tuning theory than knowing the software.


Thanks once again!

Ryan: I think its a great time saver to use the average tuning file for a pitch raise. I thought using a predetermined curve would have no use, but as a rough means to get everything in place is a good idea.

I´ll also say I share with you the idea of measuring Cs and Fs (I use those very same notes) and also at the break given that a sudden change in Ih occurs. This gives a better Rx of the piano