common_voice_ug/
├── mcv-scripted-ug-v24.0.tar.gz # 原始压缩包 (9.7GB)
└── cv-corpus-24.0-2025-12-05/
└── ug/ # 维吾尔语数据
├── clips/ # MP3音频文件 (11GB, 305,081个文件)
├── train.tsv # 训练集 (122,911条)
├── dev.tsv # 开发集 (14,656条)
├── test.tsv # 测试集 (14,656条)
├── validated.tsv # 已验证数据 (269,390条)
├── validated_sentences.tsv # 句子库 (231,426条)
├── clip_durations.tsv # 音频时长 (305,081条)
├── invalidated.tsv # 无效数据 (21,500条)
├── other.tsv # 其他数据 (14,191条)
├── reported.tsv # 举报数据 (1,539条)
└── unvalidated_sentences.tsv # 未验证句子 (11,032条)
common_voice_ug_{数字ID}.mp3
例如: common_voice_ug_38668292.mp3
client_id: 0432a01b98294eb29faf664bb81eb0340a2a5645...
audio: common_voice_ug_38668292.mp3
text: بۇ چوقۇم بىر ئالدامچىلىق، ئالدىرىما، مەن بېرىپ قاراپ باقاي.
(这一定是个骗局,别急,我去看看。)
up_votes: 2
down_votes: 1
age: (未标注)
gender: (未标注)
client_id: 04455100eeff4b339507fd06d6bc4fff5f6a4f3da0d1e8bb662fb8541425d7a9...
audio: common_voice_ug_39704025.mp3
text: ئۈچىنچى ئادەت، مۇھىم ئىشنى ئالدىدا قىلىش.
(第三个习惯,重要的事情先做。)
up_votes: 2
down_votes: 0
age: (未标注)
gender: (未标注)
client_id: 12b431b35706af75277461b62923c2d758349e2d0c96e36760e1532e0f583c96...
audio: common_voice_ug_36990048.mp3
text: بىر نەرسىگە پۈتۈن ۋۇجۇدۇڭ بىلەن بېرىلسەڭ پۈتۈن دۇنيا ساڭا ياردەم بېرىدۇ.
(如果你全身心投入一件事,全世界都会帮助你。)
up_votes: 2
down_votes: 0
age: (未标注)
gender: (未标注)
1# 安装huggingface-hub
2pip install huggingface-hub
3
4# 下载整个仓库
5huggingface-cli download anke01/common-voice-uyghur-v24 --local-dir ./common-voice-uyghur-v24
6
7# 或仅下载压缩包
8wget https://huggingface.co/anke01/common-voice-uyghur-v24/resolve/main/mcv-scripted-ug-v24.0.tar.gz
1import pandas as pd
2
3# 读取训练集
4train_df = pd.read_csv('cv-corpus-24.0-2025-12-05/ug/train.tsv', sep='\t')
5
6# 读取验证集
7validated_df = pd.read_csv('cv-corpus-24.0-2025-12-05/ug/validated.tsv', sep='\t')
8
9print(f"训练集大小: {len(train_df)}")
10print(f"已验证数据大小: {len(validated_df)}")
1@misc{commonvoice2025,
2 title = {Common Voice Corpus 24.0},
3 author = {{Mozilla Foundation}},
4 year = {2025},
5 month = {12},
6 howpublished = {\url{https://commonvoice.mozilla.org/}},
7 note = {Accessed: 2026-02-28}
8}