Qwen3-ASR
Qwen3-ASR Overview
STT (Speech-to-Text) AI-powered speech recognition.
Qwen3-ASR is Alibaba Cloud's latest open-source speech recognition series built on the Qwen3-Omni foundation.
- Multilingual: Supports 52 languages and dialects with Auto-Language Identification (LID).
- Robust: Reliable performance in noisy settings, music, or long audio (up to 20m).
- Versatile: Supports both real-time streaming and offline batch processing.
- Lineup: Includes 1.7B (High-perf), 0.6B (Efficient), and ForcedAligner models.