A phoneme recognition framework based on auditory spectro-temporal receptive fields.