摘要 |
<p>PROBLEM TO BE SOLVED: To attain a natural dialog by recognizing an operating state from the voices, gestures, etc., of a user, detecting the rhythm of the user's speech and controlling the output. SOLUTION: The voice pitch of the input data (I1 ) undergoes the A/D conversion, a voice pitch level is recognized in every prescribed processing unit, and the rise pitch of a prescribed level is recognized. These recognized level and pitch are sent to a rhythm detection means 3. The means 3 detects a cycle (rhythm) from the voice pitch of a user and also detects the dialog rhythm of the user together with the time information. A dialog management means 4 estimates the timing for the 'response-giving' answers and the speech switching answers which are outputted from a computer in response to the user's rhythm. Then an output means 5 outputs the 'response-giving' answers through an action where a human being model synthesized by the CG(computer graphics) shakes vertically his head to show the voice output 'yes'.</p> |