A New Approach for Formants Emphasis Using Fuzzy Techniques

In this paper, we propose a new method of emphasis the formants extracted from spoken sequences. Two fuzzy systems are used: the first performing the detection of the formants based on a fuzzy concatenation of several spectral envelopes (smoothed spectrums) and the second performing the accentuation of the local maxima corresponding to the formants. The purpose of this method is to improve the recognition rate of emotions, the fundamental frequency F0 and formants F1-F4 being the main carriers of information with an emotional charge. Another utility could be the better auditory perception of emotions in sound files with emphasized formants, especially in environments with high ambient noise.