| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a voice interpretation system in which both of real time processing and batch processing are performed by accurate voice recognition of a voice signal which is once CODEC converted by a voice signal recording device such as an IC recorder. <P>SOLUTION: A voice recognition device 220 utilizes a monitor function of the voice signal recording device 210 ahead of the voice recognition, and using the voice signal which is contracted and expanded by a CODEC conversion section 212, a training operation for extracting a characteristic data of human voice to be recognized is performed in real time and the extracted characteristic data is recorded as a training data. Then, the voice recognition device 220 converts the voice signal from the CODEC conversion section 212 to a text data based on the training data. Thereafter, the text data obtained by the voice recognition by the interpretation device 230 is interpreted, and the text data obtained by the interpretation is voice synthesized by a voice synthesizer 240 and the obtained voice signal is regenerated as voice. <P>COPYRIGHT: (C)2007,JPO&INPIT |
