CosyVoice Alibaba Group

CosyVoice is an AI-powered voice input app from Alibaba Tongyi that turns speech into polished text. It provides real-time high-quality speech-to-text that strips filler words and repetitions, recognizes spoken self-corrections such as "no wait, change that to", auto-formats rambling speech into numbered lists, tables, or outlines, and drafts complete emails or meeting invites from a single spoken instruction. It understands Shanghainese, Cantonese, Sichuanese and more, transcribing them into standard Mandarin, and offers multilingual translation, smart speaker separation, smart summaries, and smart to-do tracking.