Optimum estimation and detection of distortion product otoacoustic emissions in white Gaussian noise

Describes the estimation and detection of distort product otoacoustic emissions (DPOAEs) in presence of white Gaussian noise. The powerful maximum likelihood (ML) estimation and generalized likelihood ratio test (GLRT) are derived for this problem. The resulting optimal estimator and detector are shown to be superior over the conventional approaches. It is also proven that the conventional unwindowed FFT coincides with the optimal method, under the constraint that the stimulus tones are properly adjusted to lie within the FFT bins.