BVCC
收藏arXiv2025-09-30 收录
下载链接:
https://codalab.lisn.upsaclay.fr/competitions/695
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是一个大规模的平均意见得分(MOS)数据集,它包含了来自以往暴雪挑战赛和语音转换挑战赛的音频样本。每个音频样本都由8位听众进行评分,且验证集和测试集的听众是不同的。该数据集规模庞大,包含了训练集、验证集和测试集,分别包含4974、1066和1066个音频样本。任务目标是自动预测平均意见得分(MOS)。
This is a large-scale Mean Opinion Score (MOS) dataset containing audio samples from previous Blizzard Challenges and Voice Conversion Challenges. Each audio sample is rated by 8 listeners, and the listeners for the validation set and test set are distinct. The dataset comprises training, validation, and test splits with 4974, 1066, and 1066 audio samples respectively. The task objective is to automatically predict the Mean Opinion Score (MOS).
提供机构:
VoiceMOS Challenge



