lmms-lab/voicebench

VoiceBench数据集是一个用于评估基于大型语言模型的语音助手的基准数据集。它包含了多种配置，如advbench、alpacaeval、bbh等，每种配置都有不同的音频和文本特征，以及测试集的样本数量和大小。数据集支持多种语言，包括英语、中文等，并且提供了数据文件下载路径和许可证信息。

VoiceBench is a benchmark dataset for evaluating LLM-based voice assistants. It includes various configurations such as advbench, alpacaeval, bbh, etc., each with different audio and text features, as well as the number of samples and size of the test set. The dataset supports multiple languages, including English, Chinese, etc., and provides data file download paths and license information.