| 摘要 |
PURPOSE:To obtain a voice synthesizer for vocal sound inexpensive and with high quality by generating a pitch pattern for vocal sound synthesis by converting a code pattern to a pitch pattern with a table showing relation between the scale (code) and pitch. CONSTITUTION:A decode pattern is converted to the pitch pattern for vocal sound with a table 2 for pitch data decoding showing the relation between the code i.e. the scale and the pitch. At such a case, the pitch pattern of synthetic voice takes the value of the pitch at a representative point given to a more in all the points in one mora, and it is generated by performing the linear interpolation of the pitch of the mora in the neighborhood of the mora and the transition part of the mora. Also, the data of height in audio data is set as the scale of the representative point, and is set as the scale encoded with a code for which a numeral is allocated sequentially to each scale. The rule synthesis part of a signal processing circuit 1 inputs the code, and generates the code pattern on the code, then, performs the linear interpolation. In such a way, the voice synthesizer for vocal sound inexpensive and of high quality can be obtained. |
