Alibaba's Qwen Launches Qwen-Audio 3.1 Voice Models with Up to 95% Price Cut

Alibaba's Qwen has launched the Qwen-Audio 3.1 series of voice models, releasing five new models covering speech recognition, synthesis, real-time interaction, and audio creation. The ASR model supports 30 languages and 16 Chinese dialects with a first-token response of about 160 milliseconds. The TTS model enables natural cross-language voice synthesis. The Realtime model supports full-duplex interaction, allowing users to interrupt anytime and call Agent tools.

The new TTS-Next model can generate voices, sound effects, and ambient sounds together. ASR-Next understands emotions, environmental sounds, and mechanical sounds, and can describe audio events.

Prices have been cut significantly across the line: TTS down about 70%, Realtime down about 85%, and ASR down 95%. The models are being integrated with Qwen Office Agent and Qwen AI glasses.