[pianotech] Tuning test scoring

[David Renaud]

Hello

  Having conducted a number of exams now, lets see if this can help......
  I ll address your numbers as well as some broader issues this raises.

  Cybertuner will generate a tuning stretch based on samples it listens
to and calculate the "best" smooth stretch curve according to its programing and the choice of stretch number. This will provide a nice
tuning that would be a good basis as a starting point from which to launch
refinement into a "master" tuning. During master tuning one CTE and two RPT
listen to the interval relationships and smooth out the progressions and 
"balance" the interval relationships. 

   The result is a master tuning that will follow details of tone
in the given pianos string scaling, age, and quirks that deviates from
the smooth curve of a program like cybertuner or tunelab. There will be
visual jumps that deviate gradually away from the smooth curve through
a string size and jump back to the baseline curve at string size changes.

    These jumps, or deviations from a theoretical curve can be more then
1 cent in the temperament of the piano, and much more outside the temperament. Scoring perfect calculated curves against prov en master 
tunings giving good test results can actually generate some errors.
Although a perfect computer tuning zeroed will score very high, it is 
not perfect up against the master. A master will yield more detail according to individual notes and quirks in a given piano.

   All that to say simply; not knowing the piano, and not having a real
master tuning on the piano you are tuning the numbers don't give a real mark. The master tuning itself might have significant variations from
the smooth curve calculated by a program. These variations are justified by string scaling,and justified by interval progression demands verified
with careful aural listening.

   That said, it looks like your in the ballpark. If these numbers were real against a real master take each difference of over .9cent to add points, then for the temperament multiply the points by a multiplier of 2.5, 
subtract that from 100 to get your mark. So for exp ample if you come up with 8 points, multiply by 2.5= 20% deduction = 80% Note that a note 1.97
cent off would be 2 points, 2.98 off would be 3 points on the one note.

  You have a great tool to train yourself, if you are getting within
-.6 cents of the curve with only one or two vatiations more then 1 cent
you are come along nicely. I would still advise have an RPT listen to
the interval progression. 

  An exercise that would be VERY GOOD for all of us is to take the
errors indicated by the machine and challange them aurally. Listen 
to at least four checks related to that note in differnt ways and evaluate
if the aural checks confirm if the note is flat, sharp, or OK. If you get
contradictiory reports from a check and all others confirm the opposite, 
the proponderence of evidence wins. Be also aware that checks only work
once we are within certain minimium perameters; A very poor temperment
with yield unreliable checks. 

   Hope this helps.
   The machine is a great tool towards aural training. 
   
   How is your margin of error read on unisions and stability.
   You can test yourself perfectly on these and generate real marks.

                                      David Renaud
                                      Canada
 
     





  
 






      __________________________________________________________________
Yahoo! Canada Toolbar: Search from anywhere on the web, and bookmark your favourite sites. Download it now at
http://ca.toolbar.yahoo.com.