rendchevi/emilia-yodas-english-neucodec-VJKL-25k-prep

该数据集是一个用于语音或文本处理任务的结构化数据集，包含多个特征，如文本内容、语言标识、说话人信息、音素序列、DNSMOS音频质量评分、音频时长、电话计数、编码列表和说话人嵌入向量。数据集被划分为训练集（25,000个样本）、评估集（2,500个样本）和测试集（25,000个样本），总大小约为317MB，下载大小约为306MB。这些特征支持语音合成、语音识别、说话人识别或音频质量评估等应用。

This dataset is a structured dataset for speech or text processing tasks, featuring elements such as text content, language identifier, speaker information, phoneme sequences, DNSMOS audio quality scores, audio duration, phone counts, code lists, and speaker embedding vectors. It is divided into training set (25,000 examples), evaluation set (2,500 examples), and test set (25,000 examples), with a total size of approximately 317MB and a download size of about 306MB. These features support applications like speech synthesis, speech recognition, speaker identification, or audio quality assessment.