SPEECH-COCO

该数据集基于COCO数据集，通过Voxygen的文本转语音系统合成了包含多种声音和调整后的语速的音频。该数据集包含了八种不同的英语发音，并引入了语言不流畅性。其任务是进行音频标注。

Based on the COCO dataset, this dataset synthesizes audio samples via Voxygen's text-to-speech system, which features diverse voice types and modified speech speeds. It includes eight distinct English accents and introduces linguistic disfluencies. The core task of this dataset is audio annotation.