Sixth European Conference on Speech Communication and Technology

Budapest, Hungary
September 5-9, 1999

Detection of Non-Stationarity in Speech Signals and its Application to Time-Scaling

David Kapilow, Yannis Stylianou, Juergen Schroeter

AT&T Labs-Research, Shannon Laboratories, Florham Park, NJ, USA

This paper describes an automatic method for the detec-tion of non-stationarity in speech signals. It is based onthree measures of non-stationarity using Line Spectrum Frequencies (LSFs), the derivative of RMS values, and acombination of these two features. The application of the proposed method to time-scaling of speech signals is also presented. Results from an informal listening test sup-port its usefulness. Following these results, the method seems to be a powerful tool for the automatic control oftime-scale factors based on the characteristics of the in-put speech signal. Listeners preferred our new method over applying a constant time-scale factor in 90% of allcases. Other possible applications of the proposed toolare also discussed.

Full Paper (PDF)   Gnu-Zipped Postscript

Bibliographic reference.  Kapilow, David / Stylianou, Yannis / Schroeter, Juergen (1999): "Detection of non-stationarity in speech signals and its application to time-scaling", In EUROSPEECH'99, 2307-2310.