Qwen3-ASR

model
Alibaba

Qwen's open speech-recognition line (backfilled from January 2026; Transformers-native "-hf" ports shipped June 26): 0.6B and 1.7B models built on the Qwen3-Omni stack, covering 30 languages plus 22 Chinese dialects with unified streaming and offline decoding. Shipped alongside Qwen3-ForcedAligner-0.6B, a non-autoregressive timestamping model. Apache-2.0; over 1.3M cumulative downloads by mid-2026.

HuggingFace (1.7B)HuggingFace (0.6B)

Model Details

License Apache 2.0

Variants

Name Parameters Notes
Qwen3-ASR-0.6B 0.6B —
Qwen3-ASR-1.7B 1.7B —
speechasropen-weight

Related