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

Since it's so quick and easy to sample notes I usually just go for it and sample pretty much all 88, or most of them. There are two advantages to doing that. One is that you have plenty of samples for EPT to use and second is that you get a nice over-view of where the piano is at compared to the calculated curve.

Overpull -

Depending on how one uses the lever - slow pull, quick jerk, pointing left or right on an upright... less or more overpull is required to approach a one pass tuning. Also, as one tunes many of the same model, assumptions can be made to assist with overpull. A "plus or minus 1 or 2 or something" setting allows for a little less, or more overpull as needed - based on the experience from previous attempts.

Kindof a learning curve between tech and ETD to enhance the results.

I know the more controls included, the less "easy" the software becomes, but it is a real value to techs that want to dig a bit deeper!

Ron Koval

Hi Anthony,

Thanks a lot for sharing.

Do 12ths come out on the narrow or wide side?

Together with 12ths, have you been able to analyze also double-octaves and the RBI's beat-rate progression?

Do you use Audacity for your analysis, or?

Kind regards,

Alfredo

@Ron, Thank you, I'll see what I can do.

@Alfredo,
During the earlier development of the app I spent a lot of time looking at graphs of calculated beat rates, but it's been a while since I've done that and the graphs I was looking at wouldn't be of much help. (They were pretty messy and you have to stare at it for a long time to understand what is going on.) A few months ago I was doing a small presentation for my local PTG chapter and I made a graph for that that I will try to post here with a couple disclaimers. To produce this graph I tuned a single piano (Yamaha U1) five times back to back, strip-muted for single strings only, and recorded each string immediately after tuning it so that its pitch wouldn't be affected by any change to neighboring strings. The first tuning was a quick first pass to get the piano close. For each of the next four tunings I used a different software package and tuned the piano as precisely as I could to what that software said was right (stopped strobe bars, spinners, full blush, etc.). Then I analyzed the recordings, running all four through the same script in Matlab to extract the frequencies of each harmonic. I used those frequencies to calculate the 12th beat rates. Those are what is plotted in the graph I'm linking here. (The y-axis is the beat rate in Hz. Positive means wide, negative means narrow.)



Now for the disclaimers:
1. For all of the software packages I'm not trying to make any statements as to the quality of their tunings. This is just one graph of one interval on one piano.
2. I used each program using the default out-of-box settings or recommended settings as much as possible. TL and RCT both got to listen to the recommended measured notes, while VT and EPT both got to listen to the quick first pass tuning and then were both locked. The one exception on the recommended settings was that I turned off Smart Tune on RCT because I was recording the strings immediately after tuning and before any "settling" happened.
3. For RCT and VT I was using older pocket-PC versions of the software on an HP IPAQ, so the results may not be the same for the newer iOS versions.
4. I did my best to be scientific through the whole process, trying to eliminate any systematic biases, but I don't claim to be perfect.

My observations from this graph was that for all the software packages the 12ths started out a bit wide in the bass (monochords), dropped to near-zero for most of the mid-section, and got a bit wild in the treble as the beat rates for everything go up exponentially up there. The graph would look a lot cleaner if it were plotting cents.

Thanks!

I'm new to Android. Got an old phone from a family member to make it easier to test than my big Chromebook. Beta version ready to go.

Anthony, what is your preferred method for communication about EPT? Should I keep posting suggestions or observations here??

Ron Koval

Hi Anthony,

Is there more to your presentation and could you please share it? My interest in EPT is growing and I want to learn more about it.. I'm getting some nice tunings and enjoying the using it.

Many thanks,
Brent

Hi Ron,

Hoping you will continue to post here or on some public forum. Your posts are always enlightening.

Thanks,
Brent

Ron, in the past I've taken questions and suggestions by email, but I'm happy to continue answering questions here or wherever.

Brent, here's the other main graphic from the presentation. It's a companion to the above and was produced with the same recordings and Matlab script. The extra disclaimer on this one is that in order to display complete Railsback curves it had to extrapolate some data for missing fundamentals in the low bass by using an inharmonicity constant calculated from higher harmonics.



This was the only side-by-side test of this sort that I've done, and I found it to be more useful as a sanity check than anything else. The rest of the presentation was basically showing people the functions, explaining some of the rationale and motivation, and performing a 15-minute muteless pitch raise on a piano that was "prepared" at 20 cents flat.

Can you explain your approach to a muteless pitch correction? It didn't seem like the "peaks" graph on EPT was precise enough for that! (I'd imagine you'd have to narrow the range down to a third or so, and have a target line linked to the overpull calculation?)

Or are you able to focus on the needle only and just work by ear for the 2nd and 3rd strings?

Interesting...

Ron Koval

I’m also curious as to how you you do a muteless pitch raise. I have done it often with tunelab because you can see three separate peaks for each string. Thanks.

It definitely involves some "reading between the lines" and a good deal of practice. It works best on pianos between about 15 to 40 cents flat, and doesn't really work well for the highest treble.

Basically how it works is you tune the first string while focusing only on the strobe wheels and ignoring everything else. The strobes tend to single out whichever string is closest to the target pitch, so when one string is in tune you will have one or more of the rings stop. It may be jerking around and flashing, but it will be stopped on average. The second and third strings are then tuned by ear. The hardest part of the whole process is tuning the second string. It's partly by feel (based on how far you turned the first pin) and partly by ear, listening to one pitch sliding into another with a fast but smooth pull. The third string is easy if you get the second. If you get lost and don't know which string is where anymore, either pull out a mute or move on and fix it later during a quick touch-up before the fine tuning. If you're interested in trying this out it's easiest on the bi-chords. I suspect most people will be happy to stick with their normal process and only attempt this on birdcages (which is where I first tried it).

I have received requests for a zoomable FFT spectrum from TuneLab users, and it's on my long-term list of things to do.

What octave settings did you use for TuneLab when you compared the four programs? I'm wondering why TuneLab appears so wide in the treble compared to the others. I'm a longtime TuneLab user.

I didn't change the default, so I think 6:3 in the bass and 4:1 in the treble. I'm not sure why TL was sharper than the others up there. Perhaps it was the C5 measurement and the way it extrapolated inharmonicity for the Treble? I wouldn't read too much into it, as up there a difference of 5 cents isn't that big a deal, and without actually listening to the piano we can't objectively say which stretch is better anyway. On a different piano a different program may have the most stretch in the high treble.

So the fairly conservative stretch of 6:3 and 4:1 would explain why the 3:1 12th's appear wide, right?

Tried a pitch raise with the beta version and the click boxes to continue didn't appear in landscape mode.

For some reason, I wasn't able to get the pitch raise function to work today - put it back in portrait mode to click continue, then played through the presample notes, but don't think I actually got to the place where the software finds out how far off each note starts.. When I tried to start, the dialog popped up that not enough information was available to start. Tried twice, then moved on.

Found the tuning graph screen helpful in pointing out which notes may have shifted while tuning the unison. Needle and spinner seemed better than older version, but at that point, I was just having EPT follow along while I used the Verituner. Very close agreement on this old Story & Clark spinet between the two, with the biggest difference having EPT want the bass lower than I prefer.

Ron Koval

Ron,
Thanks for that feedback. I'll make sure this bug gets fixed in the next release. It sounds like there's an issue with screen resolution (thus the continue button not showing up in landscape) combined with Android's nasty habit of restarting/deleting everything every time you rotate the screen.

Ah, didn't know about that Android thing!

Just came back from a Yamaha p22 tuning - overpull was flawless this time - as a matter of fact, the tuning came out really well - all the way to the bottom.

As I've spent more time with the spinner, there is a lag - similar to OnlyPure, but not that much. Probably tuning just past the attack rather than at the attack. Once I've gotten the hang of it, works so much better! Still tough on Bb7-C8 getting a solid read.

Nice work!

Ron Koval

I haven't used the pitch raise function on a piano yet but I could see pitch raise function boxes when moving from portrait to landscape mode just now while experimenting.

Here's a question:

In the tuning file section, there is a place to note the lowest plain-wire string - as well as choose the size and type of piano.

I assume the note choice has an effect on pitch raise, does the size and type of piano also have any effect on tuning calculation or pitch raise function? If so, there should probably be a warning dialog to encourage the user to fill in those choices before starting tuning.

Ron Koval

The lowest unwound string does not affect the tuning calculation. It does have a very small effect on the pitch raise, slightly lowering the overpull percentage for the wound strings. I'm currently working on a better model for the overpull percentage, and when that comes out then it will actually ask you for the lowest unwound string at the beginning of the pitch raise setup.

The "piano type" is only for file management purposes, to give you a pretty picture to remember what type of piano you are opening.