摘要 |
<p><P>PROBLEM TO BE SOLVED: To provide a method for reproducing a voice similar to the original voice of a speaker and to facilitate communication when it is difficult to communicate with clear voices such as when supporting esophageal speech for speech disordered people or assisting communication in highly noisy environment. <P>SOLUTION: Transfer functions are generated for each subword based on input signals of conduction sound in the body of a speaker and voice data used as a model of an output voice. When the speaker makes an utterance, tone of the input signals are continuously converted for each subword by the transfer functions, and convolution operation of the tone-converted subwords is performed through a digital filter to reproduce continuous output voices. Speech support is thereby achieved with voices of tone similar to the output voices. <P>COPYRIGHT: (C)2009,JPO&INPIT</p> |