| 摘要 |
PURPOSE:To prevent learning efficiency from lowering by calculating a score from a synthesized voice and a learning voice, and re-registering a voice pattern for learning vocalized when the score exceeds a threshold value on a voice dictionary. CONSTITUTION:A voice synthesis part 4 generates a voice parameter from the voice pattern of the voice registered on the voice dictionary 2 which becomes reference, and the voice parameter of the phonetic symbol of a word desired to learn is predicated and extracted at a parameter prediction/extraction part 6 based on a generated voice parameter, then, it is used. A generated synthesized voice is provided for a learner, and simultaneously, it is inputted to a score calculation part 8, and the score is calculated from the voice pattern vocalized by the learned and a distance, and when the score exceeds a certain threshold value, the score calculation part 8 sends a specific control signal B9 to a voice registration part 3, and an operation to re-register the vocalized voice pattern of the learner at that time is performed. |
