TTS dataset

该数据集包含了在专业语音制作工作室录制的40,244条语音，由一位女性发音人完成，总时长约40小时。这些音频样本被用于训练多速率注意力架构的文本转语音系统，同时采用了一种无监督的对齐算法来提取特征。该数据集的规模为40,244条语音，任务是对文本转语音（TTS）的合成。

This dataset contains 40,244 speech utterances recorded by a single female speaker in a professional audio production studio, with a total duration of approximately 40 hours. These audio samples are used to train text-to-speech (TTS) systems with multi-rate attention architectures, and an unsupervised alignment algorithm is adopted for feature extraction. The dataset comprises 40,244 speech utterances, and the dedicated task is text-to-speech (TTS) synthesis.