AI model that natively processes audio as the input modality to fully utilize all the information available in the audio signal.