was produced by incrementally varying parameters for bow velocity and bow force.
Vicarious Synthesizers: Listening for Timbre (ISMA98)
A tone (C3) bowed on the celletto for one and a half seconds was spectrally tracked to derive bow motion. At a 30 Hz. update rate, a physical model of the bowed string produced a replica from automatically-produced envelopes for bow velocity and bow force.
The tracking technique relies on FFT-based spectral matching where a
short segment of input (a few periods) is compared to all tones in a large
matrix produced by the synthesizer. Using CCRMA's Lisp-based CLM language
for the synthesis for the present study, a 64 x 64 matrix of tones with
various waveforms like
was produced by incrementally varying parameters for bow velocity and
bow force.
A surface formed by plotting spectral similarity to the incoming reference tone which is located in the yellow swatch, lower left of mid-point (red color is the best match).
A look at amplitudes across the surface shows the familiar correlation of loudness and increase in these parameters. RMS amplitude grows with greater velocity if a balance with force is maintained (greater amplitude is colored blue).
Schumacher and Woodhouse's successful mapping of timbral spaces provided the context for the present approach. Please see: