Bhattacharyya distance between the formants structure for robust pronunciation errors detection

In stead of tradition formant, intervals of formant structure are used to detect the pronunciation errors of vowels. The algorithm can remove both multiplicative and linear transformational distortion from speech theoretically by using the Bhattacharyya distance to evaluate the embedded relationship between statistical GMM models that represent the formant pattern of vowels. It means that the interval of formants are represented without being affected by any static distortion inevitably involved in production, encoding, transmission, decoding, and hearing processes, such as differences in vocal tract length, gender, age, microphone, room, line, auditory characteristics and so on, because of the affine invariant in Euclidean space of Bhattacharyya distance. The experiment results show that the proposed algorithm is able to be applied to pronunciation assessment effectively and reliably, where the errors of pronunciation is detected without using formant models directly.