WIDesigner release 2.4.0 now available

Oxymorons "R" Us
Forum rules
Please see forum announcement for description and rules!
User avatar
MadmanWithaWhistle
Posts: 697
Joined: Fri Jan 27, 2012 10:15 pm
antispam: No
Please enter the next number in sequence: 8
Tell us something.: Instrument maker and researcher exploring new methods of creating traditional instruments with longevity aforethought. Player of the whistle, flute, and continental European border pipes.

Re: WIDesigner release 2.4.0 now available

Post by MadmanWithaWhistle »

Trying to get back at whistlemaking, but encountered some additional questions:

1. Does "Blowing level" affect anything beyond the position of the blue circles on the graph? It doesn't affect the values in the table.
2. Is there any benefit to using "DIRECT" calculation? It takes marginally longer but other than that I haven't seen a change in the values.
3. Is it a problem to configure a tuning file with only the notes, fingerings, ratios, and frequencies of your desired scale? I calculated my desired temperament from Hans Bracker's "trad" frequencies, which doesn't include unused values like minor seconds etc.
4. Is "1 whistle calibration tool" intended to adjust hole spacings? T3 got moved by a thou when I ran the tool.
5. Something has gone badly awry with "4. hole and spacing optimizer." When I use it according to the instructions, it gives me a length of 15 inches over the original Eb tube's 10.

As always, procedure notes are here: https://docs.google.com/document/d/1zP4 ... sp=sharing
And the tuning file in question is here: https://drive.google.com/file/d/1-2aALn ... sp=sharing

I appreciate any recommendations here!

The headjoint's done, I just need to calculate some decent body tube dimensions and I can release these.
Tunborough
Posts: 1419
Joined: Sun Dec 05, 2010 2:59 pm
antispam: No
Please enter the next number in sequence: 10
Location: Southwestern Ontario

Re: WIDesigner release 2.4.0 now available

Post by Tunborough »

WIDesigner now has a discussion board. For other users who might be interested, I'd appreciate if you could post these questions there, or with your permission, I can post them.

1. On the graph, the Blowing Level setting affects only the position of the blue circles. These mark the predicted playing frequencies, so in the table, the Blowing Level affects the predicted playing frequency (and the deviation).

2. Depends on where you start, and what you are optimizing. For most purposes, optimizing the hole layout from a plausible starting point (like another D whistle), DIRECT is (to be blunt) a waste of time. If you're optimizing a lot of interacting dimensions, such as hole layout and bore profile, DIRECT is less likely to get caught in a local optimum and miss a much better global optimum. If you're starting a long way from your target, such as starting from a D whistle layout to create an A whistle, DIRECT will explore a broader range of possible solutions.

3. Not a problem. The tuning file should have only the notes you want to optimize for.

4. WIDesigner does its internal calculations in metres. If the whistle calibration optimizer changed a hole position measured in inches, it was likely round-off error in the conversion from inches to metres and back.

5. (I'll need to take a closer look. It may be a couple of days before I get to it.)
Tunborough
Posts: 1419
Joined: Sun Dec 05, 2010 2:59 pm
antispam: No
Please enter the next number in sequence: 10
Location: Southwestern Ontario

Re: WIDesigner release 2.4.0 now available

Post by Tunborough »

5. Can you elaborate on point 3f in your notes? That would be the problem.
If (target) “Frequency” column has data, HOLE optimization functions will not work. Delete anything here and just put the min/max when using HOLE Optimization, but required for calibration.
If min/max frequencies are provided for a note, the calibrator ignores the Frequency column; otherwise, it assumes the Frequency column contains the actual measured playing frequency and calibrates for that.

For other optimizers, the target frequencies are what optimizers is working toward. (They ignore the min/max frequencies.) If you don't give the optimizer any target frequencies, you're telling it to go, but not telling it where; the results are unpredictable, and could include a wildly unexpected bore length (point 4a in your notes). The fact that WIDesigner let you do this at all is an issue we should look at fixing. Once you got the 16" bore length, if you then tried optimizing to an Eb5 tuning with target frequencies, the optimizer could head for the wrong local optimum, and end up with Eb5 as the second harmonic of the instrument's bell note (around 21" bore length, point 4b in your notes).
User avatar
MadmanWithaWhistle
Posts: 697
Joined: Fri Jan 27, 2012 10:15 pm
antispam: No
Please enter the next number in sequence: 8
Tell us something.: Instrument maker and researcher exploring new methods of creating traditional instruments with longevity aforethought. Player of the whistle, flute, and continental European border pipes.

Re: WIDesigner release 2.4.0 now available

Post by MadmanWithaWhistle »

Thanks again for helping me understand how the different attributes are used, and pointing out the rounding! Is there a "local cache" or anything of that nature that can be purged? I think some of my less-reproducible anomalies might stem from something like that. I created a github account and am going to make a new format for my usage notes that will hopefully be a little cleaner. Once I get the hang of the tool I am excited to use it to optimize a Pastoral Pipe chanter as part of a wee competition I have going.

Should my tuning file always contain target frequency, min, and max of the most recent whistle, then, since the optimizers seem to just take what they need?

Edit: new usage notes are here; much easier on the eyes: https://docs.google.com/presentation/d/ ... sp=sharing

Edit2: still getting a 21 inch whistle, even on the FeadogMk1 test files (feel free to continue the discussion on the github thread)
Post Reply