This paper describes several methods for reducing the enormous amount of memory in digital pianos based on sampling techniques. We briefly describe the synthesis and sampling techniques as methods of sound generation.
In order to reduce the memory requirements while not compromising sound fidelity, a total of 88 keys are divided into 14 groups according to the grouping methodology of commercial acoustic piano strings. A technique for reducing memory size is presented which exploits the periodic nature for eh sustaining portion of each sound waveform to be reconstructed. Also, DPCM and transform coding techniques are explained in detail, the latter making full use of the masking properties of human ears.
From the experimental results, we can hardly find any noticeable difference in fidelity between the original and reconstructed sound.