An odd-DFT based approach to time-scale expansion of audio signals

A new time-scale expansion algorithm based on a frequency-scale modification approach combined with time interpolation is presented. The algorithm is noniterative and is constrained to a blind modification of the magnitudes and phases of the relevant spectral components of the signal, on a frame-by-frame basis. The resulting advantages and limitations are discussed. A few simplified models for signal analysis/synthesis are developed, the most critical of which concern phase and frequency estimation beyond the frequency resolution of the filterbank. The structure of the algorithm is described and its performance is illustrated with both synthetic and natural audio signals.