A method, computer program product, and computing system for defining model representative of a plurality of acoustic variations to a speech signal, thus defining a plurality of time-varying spectral modifications. The plurality of time-varying spectral modifications may be applied to a plurality of feature coefficients of a target domain of a reference signal, thus generating a plurality of time-varying spectrally-augmented feature coefficients of the reference signal.