Abstract
Typical speech wave forms are well modeled as slowly time‐varying, or so‐called locally stationary, stochastic processes. This talk outlines recent work in detecting and modeling locally stationary speech time series, and describes a new method of adaptive short‐time Fourier analysis and reconstruction based on local measures of time‐frequency concentration. While adaptive analysis measures have previously been proposed in order to overcome the limitations of fixed‐resolution schemes, the scheme presented here derives from quantifiable and rigorous notions of local stationarity. This yields demonstrable robustness properties for the case of noisy speech, as well as improved mean‐square error estimation properties and other quality improvements relative to standard estimation procedures in which time‐frequency resolution is fixed. [Work supported in part by DARPA and NSF.]
Sign-in/Register to access full text options 
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 callDisclaimer: 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.
