Rocket - Think it. Type it. Launch it.
Mini toolsQwen3 ForcedAligner 0.6B
By Alibaba
Model family: Qwen
Qwen3-ForcedAligner-0.6B is a non-autoregressive forced alignment model paired with Qwen3-ASR. It takes speech and transcripts and returns word- or character-level timestamps for clips up to about 5 minutes in 11 languages, and benchmark results show lower alignment error than tools such as WhisperX and other end-to-end forced alignment baselines. It can be used directly or via the qwen-asr toolkit and vLLM backends.
New
Audio
Gen 4
Overview
Multilingual forced alignment model that aligns speech and transcripts in 11 languages, predicting timestamps for arbitrary units in up to 5 minutes of audio with accuracy surpassing previous end-to-end aligners.
About Alibaba
Chinese e-commerce and cloud leader behind Taobao, Tmall, and Alipay.
View Company ProfileTools using Qwen3 ForcedAligner 0.6B
No tools found for this model yet.
Last updated: February 25, 2026
