摘要 |
<p>PURPOSE: To provide a speaking person confirming system having a low error rate and minimizing a confirm processing time and power for computer processing. CONSTITUTION: This system is provided with a means processing a spoken text to make it the frame of a speech, a means registering the speaking person in the system, a means 2 pulling out the speaking test frame of the speaking persons insisting to be a registered speaking person by a key board 1, etc., a distance deciding means deciding a confirming distance between this test phrase and the corresponding 'word' of a specific registered speaking person, which is inputted in a registered time, a means 7 for deciding a confirming source from these confirming distance data like and deciding whether the speaking person is the registered speaking person by combining them and a means decided by an adjacent Euclidean distance nearest between the frame of a speech related with a spoken test phrase and the corresponding frame of a speech related with 'word' inputted by the registered speaking person to the system in a registering period.</p> |