This model transforms text prompts into diverse audio outputs, encompassing multiple languages and various sounds such as speech and nonverbal cues like laughter or sighs. Designed for research, it provides high-quality, multilingual audio generation capabilities.