Fig. 1From: Feature compensation based on the normalization of vocal tract length for the improvement of emotion-affected speech recognitionThe general structure of the proposed EASR systemBack to article page