Details
- Qwen announced Qwen-Audio-3.1, upgrading its automatic speech recognition, text-to-speech, and real-time voice-interaction models.
- The release adds two models: TTS-Next for audio creation and ASR-Next for advanced audio understanding.
- The five-model lineup covers speech understanding, audio generation, real-time interaction, and creative audio production.
- Qwen said TTS prices will fall by about 70%, Realtime by about 85%, and ASR by as much as 95%.
- The lineup is intended to provide developers with a broader audio stack through Qwen’s platform rather than separate tools for each function.
- The announcement positions lower API costs as a way to reduce the expense of deploying speech and audio applications.
- ASR-Next availability was reported as forthcoming, while the broader Qwen-Audio APIs were listed on the Qwen platform.
Impact
The release combines capability expansion with unusually large stated price reductions, putting pressure on competing speech APIs to justify higher costs or match broader functionality. Lower inference prices could accelerate voice-agent, transcription, and audio-generation deployments, particularly for cost-sensitive developers. Its competitive effect will depend on latency, language coverage, reliability, and access to the new models, not pricing alone.
