maitrix-org/Voila-Benchmark

Voila数据集包含音频和文本信息，用于训练和评估Voila模型，一种能够实现实时自主交互和语音角色扮演的音频语言基础模型。数据集包括问题音频、问题文本、参考答案音频、参考答案文本、领域和来源等信息。同时提供了Voila Benchmark和Voila Voice Library两个附加数据集，分别用于模型性能评估和提供预构建的可定制化语音。

The Voila dataset includes audio and text information for training and evaluating the Voila model, a voice-language foundation model capable of real-time autonomous interaction and voice roleplay. The dataset comprises question audio, question text, reference answer audio, reference answer text, domain, and source information. Additionally, two supplementary datasets, Voila Benchmark and Voila Voice Library, are provided for model performance evaluation and offering millions of pre-built customizable voices, respectively.