Musical Background

1
Matlab Aided Piano Voicing
Team Shank Shannon
Hughes, Andrew Swick, Darius Roberts and Douglas
Duncan
Abstract In this project, we have applied
Matlabs signal processing capabilities to the
creation of a set of tools to assist piano
technicians with voicing. Voicing is the time
consuming process of adjusting the hammers of a
piano so as to achieve the most pleasing
distribution of power in the harmonics of the
note.

• Results
• Successful analysis of signal yielding
• - power in each harmonic
• - frequency in each harmonic
• - inharmonicities in each harmonic
• Successful creation of a reference library
• Prediction of voicing techniques necessary

• Objective
• To develop a tool to assist the piano technician
  with voicing which would
• - Allow the technician to store information
  about individual pianos and voicing techniques
  
• - Assist with determining what action to
  take

Musical Background - Pianos are extremely
complicated (over 9,000 pieces) - Highly
certified piano technicians can take days to
prepare a concert piano using subjective
judgments of sound quality. - The sound of the
piano is determined by its tuning (pitch),
regulation (key weight and responsiveness), and
voicing (quality of the timbre). - The voicing
is adjusted by filing, needling, and chemically
treating the felt on the hammer which strikes the
string as well as warping the hammer itself. -
Voicing is the most painstaking part of the
process, since each tweaking requires removing
the action the complex mechanism which
incorporates all the hammers. - Proper voicing
also sets the degree of inharmonicity.
Inharmonicity refers to the tendency of the
partial harmonics of a note to have slightly
higher frequencies than the exact multiples of
the fundamental one would expect. - The
process is further complicated by the fact that
no two pianos will be voiced exactly the same
way. - Software exists to assist with most
aspects of preparing a piano, but no such tool
exists for voicing.
Piano Hammer Mechanism

• Methods
• - High precision recording of unvoiced piano
  notes into Matlab
• Analysis of signals spectrum in Matlab
• - calculation of power in harmonics
• - measurement of inharmonicity
• - Usage of observed data to calculate action of
  voicing technique on the distribution of power in
  the harmonics
• Storage of this information in a permanent
  reference library
• Prediction of the effect of each voicing
  technique on a given input note
• Comparison by inner products of each result with
  ideal spectrum (different for each piano)
• Suggestion of technique on new note to attain
  best match with specified ideal

• Potential Benefits
• Could cut time required to voice a piano by
  hours
• Has use as a tool for research involving voicing
• Can assist older technicians whose high
  frequency hearing is deteriorating
• A first step towards eliminating dependence on
  human hearing for the voicing process

Conclusion Achieved successful design of
computational tool that solves our original
problem.