Abstract

Most of the pitch detection schemes for voiced segments of speech measure the pitch period as an integer multiple of the sampling period. Averaging techniques on more than one frame, for smoothing purposes, will result in pitch periods which are noninteger multiples of the sampling period. The same result is expected when interpolation takes place by changing the pitch several times within the same frame. Finally, a similar effect is observed, even without smoothing or interpolation, in those pitch detection schemes that use averaging of distances within the same frame to calculate the final pitch period for the frame. The algorithm described here permits the synthesis of male, female, and children's voices using noninteger pitches. The impact of this technique on current LPC implementations is marginal at low pitch frequencies while, at higher ones, the improvement is detectable. It is believed that as other sources of error in LPC systems are reduced, this technique will prove to be useful in improving synthetic speech quality. This paper deals with two items. It discusses the effects of truncation and rounding of noninteger pitch periods and describes the design of a pulse driving function generator, whose pitch period is a noninteger multiple of a fixed sampling period.

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

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.