摘要 |
<p>PURPOSE:To make more natural conversations by detecting an input voice level and sending an agreeable response at the time when the level lower than a certain threshold continues for a certain time. CONSTITUTION:When detecting a silent part, a silence detecting part 101 sends a signal to an interactive processing part 105. Meanwhile, a voice input part 102 sends an audio signal to a level detecting part 103 at the time of input of voice and sends the analysis result of voice to the processing part 105. The level detecting part 103 obtains a maximum value of the power of voice in each section and sends the result to a discriminating part 104. The discriminating part 104 compares this maximum value with a threshold; and when the input voice level is lower than the threshold over a preliminarily determined time, the discriminating part 104 sends the signal to the processing part 105. When receiving the signal from the detecting part 101 and the discriminating part 104, the processing part 105 sends the signal to send an agreeable response to an audio response part 106. When receiving the signal from the processing part 105, the response part 106 sends the agreeable response.</p> |