On the fusion of prosody, voice spectrum and face features for multimodal person verification | Publicación