Research map / Large language models / Language, safety and interpretability
WenetSpeech-Yue: A Large-Scale Cantonese Speech Corpus with Multi-dimensional Annotation
Longhao Li, Zhao Guo, Hongjie Chen, Yuhang Dai, Ziyu Zhang, Hongfei Xue, Tianlun Zuo, Chengyou Wang, Shuiyuan Wang, Xin Xu, Hui Bu, Jie Li and 6 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add stage 2 fine-tuning on high-confidence labels (ablation) | Whisper-medium-Yue · WSYue-ASR-eval Short | MER (%): 7.27 -> 5.05 | Improved |
| add stage 2 fine-tuning on high-confidence labels (ablation) | U2pp-Conformer-LLM-Yue · WSYue-ASR-eval Long | MER (%): 10.75 -> 7.91 | Improved |
| add stage 2 fine-tuning on high-confidence labels (ablation) | U2pp-Conformer-Yue · WSYue-ASR-eval Long | MER (%): 12.01 -> 8.89 | Improved |
| fine-tune Whisper-medium on WenetSpeech-Yue | Whisper-medium · WSYue-ASR-eval Short | MER (%): 80.82 -> 5.05 | Improved |
| fine-tune SenseVoice-small on WenetSpeech-Yue | SenseVoice-small · WSYue-ASR-eval Short | MER (%): 6.69 -> 5.23 | Improved |
| fine-tune SenseVoice-small on WenetSpeech-Yue | SenseVoice-small · Common Voice zh-HK | MER (%): 7.34 -> 8.68 | Hurt |
| fine-tune SenseVoice-small on WenetSpeech-Yue | SenseVoice-small · In-house Reading | MER (%): 6.52 -> 6.71 | Hurt |
| fine-tune SenseVoice-small on WenetSpeech-Yue | SenseVoice-small · Common Voice yue | MER (%): 8.05 -> 6.87 | Improved |
| fine-tune Whisper-medium on WenetSpeech-Yue | Whisper-medium · Common Voice zh-HK | MER (%): 62.50 -> 11.03 | Improved |
| add LLM (Qwen3-4B) to U2pp-Conformer-Yue | U2pp-Conformer-Yue · WSYue-ASR-eval Short | MER (%): 5.05 -> 4.73 | Improved |
| add LLM (Qwen3-4B) to U2pp-Conformer-Yue | U2pp-Conformer-Yue · In-house Reading | MER (%): 7.82 -> 6.21 | Improved |
| fine-tune Llasa-1B on WenetSpeech-Yue TTS subset | Llasa-1B · WSYue-TTS-eval Base | MER (%): 53.31 -> 10.89 | Improved |
Closest papers by meaning
- WorldSpeech: A Multilingual Speech Corpus from Around the World (NeurIPS 2026 · similarity 0.88)
- SwitchLingua: The First Large-Scale Multilingual and Multi-Ethnic Code-Switching Dataset (NeurIPS 2025 · similarity 0.87)
- IndicVoices-R: Unlocking a Massive Multilingual Multi-speaker Speech Corpus for Scaling Indian TTS (NeurIPS 2024 · similarity 0.87)
- MMSU: A Massive Multi-task Spoken Language Understanding and Reasoning Benchmark (ICLR 2026 · similarity 0.87)
- LAMA-UT: Language Agnostic Multilingual ASR Through Orthography Unification and Language-Specific Transliteration (AAAI 2025 · similarity 0.87)
- MultiTalk: Scaling Full-Duplex Speech Models to Long, Multi-Party, Bilingual Conversation (NeurIPS 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
