Zen3 Asr Aligner

--- license: apache-2.0 library_name: transformers pipeline_tag: automatic-speech-recognition language: - en - zh base_model: - Qwen/Qwen3-ForcedAligner-0.6B base_model_relation: finetune tags: - zen3 - zen3-asr - zenlm - qwen3-asr - forced-alignment - timestamps - subtitles --- # Zen3 ASR Forced Aligner Forced-alignment companion to Zen3 ASR. Produces word- and phoneme-level time alignments for transcripts (subtitle sync, dataset curation, dubbing). The config declares support for 11 languages. Derived by fine-tuning [`Qwen/Qwen3-ForcedAligner-0.6B`](https://huggingface.co/Qwen/Qwen3-ForcedAligner-0.6B) (Alibaba Cloud, Apache-2.0). - Architecture: `Qwen3ASRForConditionalGeneration` (`qwen3_asr`) - Parameters: ~0.6B - Base model: [`Qwen/Qwen3-ForcedAligner-0.6B`](https://huggingface.co/Qwen/Qwen3-ForcedAligner-0.6B) ## Weights This repository contains the model weights: `model.safetensors`, config and tokenizer files. The model uses the `qwen3_asr` architecture and loads with `transformers` (>= 4.57). It is API-compatible with the upstream base — follow the inference recipe on the base model card [`Qwen/Qwen3-ForcedAligner-0.6B`](https://huggingface.co/Qwen/Qwen3-ForcedAligner-0.6B). ## Provenance Fine-tuned from [`Qwen/Qwen3-ForcedAligner-0.6B`](https://huggingface.co/Qwen/Qwen3-ForcedAligner-0.6B) (Apache-2.0). See `NOTICE` for full attribution.

Source context: 12 downloads · 0 likes · Pipeline automatic-speech-recognition · Library transformers · Repo zenlm/zen3-asr-aligner