Musical tone generating apparatus

A musical tone generating apparatus utilizing a data compression method is disclosed. A musical tone is sampled and is converted into waveform data (S(n)). These waveform data (S(n)) are compressed into compressed dta (C(n)) by a linear predictive coding method and further by a differential quantization method. The number of bits of each of the compressed data (C(n)) is thus significantly smaller than that of the waveform data (S(n)). Thereafter, a memory (2) stores the thus compressed data corresponding to an attack portion of the tone and a selected part of a sustain portion of the tone. Hence, it is possible to remarkably reduce the memory storage. When generating a musical tone, the memory (2) reads out the compressed data (C(n)) of the attack portion and thereafter repeatedly reads out the compressed data of the selected part. The read-out compressed data are sequentially decoded into the original digital data. Thus, the whole waveform of the musical tone is reproduced.