Abstract
A high-performance method for pitch extraction is proposed for the purposes of real-time sequential speech processing that can be used in such applications as speech rate conversion systems. According to this method, autocorrelation functions of the input speech waveforms are calculated for one analyzed point in time using multiple lengths of the analysis windows, and the largest peaks of each autocorrelation function are detected within the appropriate ranges, after which the optimum pitch period is selected by weighting the candidates of the pitch period obtained by the number of windows. Such selection processing is carried out independently for each analyzed point without using such characteristics as the continuity of the fundamental frequencies of the entire speech segment. This method was applied to analysis of a large number of speech materials, including recordings made by different speakers and speech samples mixed with noise. The tests have demonstrated that the proposed method features pitch extraction potentials superior to those of the cepstrum pitch determination method and the LPC residual autocorrelation method within a wide range of fundamental frequencies and power levels. © 1999 Scripta Technica, Electron Comm Jpn Pt 3, 83(2): 67–79, 2000
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
More From: Electronics and Communications in Japan (Part III: Fundamental Electronic Science)
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.