Synthesis by linear prediction of the bamboo tone of Shakuhachi
F. M. Toyama, Toshio Takasugi · The Journal of the Acoustical Society of America · 1978
The study was concerned with synthesis of bamboo tones of shakuhachi entirely by the technique of linear predictive coding which is commonly used in speech synthesis. The bamboo pipe was modeled as a linear digital filter. In applying the linear prediction method, the noise source from the lips and the transfer function proper to the shakuhachi were separated from the bamboo tones by means of predictive coefficients. From our preliminary experiments, this noise source function was found to be quasiperiodic. For synthesizing, a pulse train on which white noise was superposed was used as the source of the sound, which was amplitude modulated by the power of the bamboo tone. This pulse train was combined with the predictive coefficients for synthesizing. The results were satisfactory, recreating its natural tone quality.