VERFAHREN UND EINRICHTUNG ZUR ANALYSE UND SYNTHESE VON MENSCHLICHER SPRACHE.

摘要

In a speech analysis and synthesis method and device, human speech is analyzed and a feature parameter (11) is extracted from the human speech. The feature parameter (11) is quantized (12) and a quantization parameter (15) is generated. Part (22) of bits of the quantization parameter (15) is a predetermined bit pattern when the feature parameter (11) is on a predetermined range. The part (29) of bits of the quantization parameter is transformed into an additive parameter (23) of at least one bit. The human speech is synthesized on the basis of the quantization parameter (15) and the additive parameter (23). With this method and device, it is possible to non-linearly quantize the human speech into the quantization parameter (15) of small bits and to generate the synthesis parameter (24) from the quantization parameter (15) by using circuit of small size.