发明名称 SPEECH ENHANCING DEVICE, METHOD, PROGRAM, AND RECORDING MEDIUM OF THE SAME
摘要 PROBLEM TO BE SOLVED: To provide a speech enhancing technique capable of reflecting the characteristic of an easily audible speech and improving clarity without requiring preliminary tuning of a control amount against noises having various characteristics.SOLUTION: A speech enhancing device calculates a power ratio of a speech spectrum to a noise average spectrum using the speech spectrum, a formant frequency, and the noise average spectrum, estimates a target power ratio which is a power ratio at which easy audibleness is achieved corresponding to a target clarity score in a formant greatly affecting the clarity in speech on the basis of a relationship between the previously acquired easy audibleness in speech and the power ratio, and includes a spectrum conversion part for converting the speech spectrum into a formant enhancement spectrum in the formant frequency on the basis of the target power ratio and the power ratio; and a speech synthesis part for synthesizing enhancement speech data s(t) on the basis of the formant enhancement spectrum.
申请公布号 JP2014102349(A) 申请公布日期 2014.06.05
申请号 JP20120253620 申请日期 2012.11.19
申请人 NIPPON TELEGR & TELEPH CORP <NTT> 发明人 KAMIYAMA HOSONA;MIZUNO HIDEYUKI
分类号 G10L21/007;G10L21/0332;G10L21/057 主分类号 G10L21/007
代理机构 代理人
主权项
地址