Alibaba Launches AI Voice Platform CosyVoiceStudio

On August 7, Alibaba announced the launch of CosyVoiceStudio, an all-in-one AI-powered voice productivity platform built on its proprietary audio foundation model, Qwen-Audio. CosyVoiceStudio comprises three core components: CosyFlow, a voice recording tool; CosyAgent, a voice-based intelligent agent; and CosyCreative, an audio content creation tool. Users can download the app by searching “CosyVoice” in the iOS, Android, Mac, and Windows app stores. A limited-time free trial is currently available. At this stage, CosyAgent and CosyCreative are being tested with enterprise users via a whitelist invitation system and will be fully opened to the public soon. CosyFlow supports speech-to-text transcription enhanced with semantic understanding and text generation capabilities. It automatically filters out filler words, organizes fragmented thoughts into clear expressions, and supports both real-time recording and offline file uploads. CosyAgent enables users to create intelligent agents through natural language prompts; these agents possess enterprise-specific knowledge, can invoke tools, and support real-time voice interaction. Compatible with both prompt-based and workflow-based configurations, CosyAgent is ideal for applications such as intelligent customer service and telemarketing. CosyCreative offers access to thousands of built-in voice styles and supports voice cloning. Users can generate conversational podcasts or multi-character audiobooks simply by uploading documents, providing web links, or inputting text directly.