Metraipolis › archive › dc_992bf58c0410

Qwen-Audio-3.0-ASR Technical Report — arXiv:2609.07549

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.07549
License (as archived)
CC0-1.0
Archived
2026-10-16T12:45:39.865Z
SHA-256 of the archived text
0c711a54ea078a12b0fc8024bdb54d0fdd6b97e9a43d0fbd93e977ae916caa61
Stable URL of this page
https://metraipolis.com/documento/dc_992bf58c0410

How to cite

Qwen-Audio-3.0-ASR Technical Report — arXiv:2609.07549 — arXiv (en), https://arxiv.org/abs/2609.07549 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_992bf58c0410 (Metraipolis archive dc_992bf58c0410, sha256 0c711a54ea078a12b0fc8024bdb54d0fdd6b97e9a43d0fbd93e977ae916caa61)
BibTeX
@misc{dc992bf58c0410,
  title = {Qwen-Audio-3.0-ASR Technical Report — arXiv:2609.07549},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.07549},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_992bf58c0410; sha256 0c711a54ea078a12b0fc8024bdb54d0fdd6b97e9a43d0fbd93e977ae916caa61},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.07549. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

In recent years, automatic speech recognition (ASR) has witnessed transformative advancements driven by three complementary paradigms: data scaling, model scaling, and deep integration with large language models (LLMs). However, bridging the gap between academic benchmark performance and real-world production utility remains a persistent challenge, particularly in handling diverse regional dialects, dynamic entities and hotwords, long-range contextual information, and disfluent spontaneous speech. In this report, we present Qwen-Audio-3.0-ASR, a Mixture-of-Experts (MoE) LLM-based ASR system designed to address these production demands through a unified, instruction-following framework. The model is built upon the Qwen backbone, and is trained on tens of millions of hours of large-scale speech data. Qwen-Audio-3.0-ASR supports transcription across 30 languages and 16 Chinese dialectal varieties spanning eight major dialect regions. Beyond multilingual and dialectal recognition, the model provides production-oriented capabilities including industry-domain entity recognition, hierarchical hotword customization, native single-pass transcription polishing, and long-audio contextual modeling. We further develop a dedicated streaming variant, Qwen-Audio-3.0-ASR-Streaming, for latency-sensitive applications. Extensive evaluations on Chinese, English, multilingual, and real-world industrial test sets demonstrate state-of-the-art or highly competitive recognition performance across a broad range of evaluation conditions, with strong performance relative to leading commercial and proprietary systems including GPT-4o Transcribe and Gemini 3.1 Pro.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-09 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.07549 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T12:30:25.124Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).