发明名称 |
CORRECTION SPEECH DETECTION DEVICE, VOICE RECOGNITION SYSTEM, CORRECTION SPEECH DETECTION METHOD, AND PROGRAM |
摘要 |
PROBLEM TO BE SOLVED: To provide a correction speech detection device capable of determining whether a correction speech (correctly retried speech) has been performed or not, not on the basis of a change in a sound feature amount for each speech.SOLUTION: A correction speech detection device of the present invention takes a sound signal as input, and presents a voice recognition result for the sound signal. The correction speech detection device takes m as an integer of two or more and includes: a reaction time measuring unit for measuring a difference between a time at which a voice recognition result for a sound signal having been input at a (m-1)th time is presented and an input time of a sound signal having been input at a mth time, as a (m-1)th reaction time; and a determination unit for determining whether a correction speech is included or not in the sound signal having been input at the mth time, based on the (m-1)th reaction time.SELECTED DRAWING: Figure 1 |
申请公布号 |
JP2016180917(A) |
申请公布日期 |
2016.10.13 |
申请号 |
JP20150061834 |
申请日期 |
2015.03.25 |
申请人 |
NIPPON TELEGR & TELEPH CORP <NTT> |
发明人 |
KAWASE TOMOKO;KOBAYASHI KAZUNORI;OMURO HITOSHI |
分类号 |
G10L15/20 |
主分类号 |
G10L15/20 |
代理机构 |
|
代理人 |
|
主权项 |
|
地址 |
|