---
title: WenetSpeech-Wu-Bench
canonical_url: "https://www.modelscope.cn/datasets/ASLP-lab/WenetSpeech-Wu-Bench"
md_url: "https://www.modelscope.cn/datasets/ASLP-lab/WenetSpeech-Wu-Bench.md"
repository: ASLP-lab/WenetSpeech-Wu-Bench
last_updated: 2026-02-09
license: apache-2.0
storage_size: "2.2 GB"
downloads: 1151
stars: 0
---

# WenetSpeech-Wu-Bench

> WenetSpeech-Wu-Bench - ASLP-lab 在 ModelScope 开源的数据集。WenetSpeech-Wu Bench

ASLP-lab/WenetSpeech-Wu-Bench 是 ModelScope 魔搭社区上的数据集，存储大小 2.2 GB，采用 apache-2.0 许可。

- **Repository**: ASLP-lab/WenetSpeech-Wu-Bench
- **License**: apache-2.0
- **Storage size**: 2.2 GB
- **Downloads**: 1151
- **Stars**: 0
- **Last updated**: 2026-02-09

Source: https://www.modelscope.cn/datasets/ASLP-lab/WenetSpeech-Wu-Bench

---

# WenetSpeech-Wu Bench


We introduce WenetSpeech-Wu-Bench, the first publicly available, manually curated benchmark for Wu dialect speech processing, covering ASR, Wu-to-Mandarin AST, speaker attributes, emotion recognition, TTS, and instruct TTS, and providing a unified platform for fair evaluation.

- **ASR:** Wu dialect ASR (9.75 hour, including Shanghainese, Suzhounese, and Mandarin code-mixed speech). Evaluated by CER.
- **Wu→Mandarin AST:** Speech translation from Wu dialects to Mandarin (3k utterances, 4.4h). Evaluated by BLEU.
- **Speaker Attributes & Emotion:** Speaker gender/age prediction and emotion recognition on Wu dialect. Evaluated by classification accuracy.
- **TTS:** Wu dialect TTS with speaker prompting (242 sentences, 12 speakers). Evaluated by speaker similarity, CER, and MOS.
- **Instruct TTS:** Instruction-following TTS with prosodic and emotional control. Evaluated by automatic accuracy and subjective MOS.