디지털 도파관 이론에 기반한 거문고 소리 합성에 관한 연구
This paper presents a sound synthesis method for the geomungo, a Korean traditional plucked-string instrument, based on physical modeling. Commuted waveguide synthesis method is used as a basic synthesis algorithm in this work. Recorded geomungo tones are analyzed by estimating frequency responses and fundamental frequency curves.
A synthesis model proposed here consists of a fractional delay allpass filter, a loss filter which combines a ripple filter and an one-pole filter, and a digital delay line. Calibration of parameters in the model is done by minimizing the difference between the magnitude response of the designed filter and estimated losses of harmonic partials.
A time-varying model which has Lagrange interpolation FIR filter as a fractional delay filter, time-varying loss filter and delay line, and a gain factor to preserve energy is proposed and evaluated to generate geomungo tones which have fluctuating pitches. Gain control using a sinusoidal function in the time-varying loss filter is also discussed.
In addition, a hybrid-modal synthesis model which synthesizes the signal with the resonators at low frequencies and with the digital waveguide model at high frequencies is proposed. Finally, real-time sound synthesis application is implemented based on The Synthesis toolkit(STK). The former shows that physical modeling based on digital waveguide theory is an e.ective method to synthesize geomungo sounds.