FunAudioLLM/SpeechFCEval

SpeechFCEval是一个基于语音的功能调用基准数据集，包含来自现有开源文本数据集和构建的智能交互场景的数据。数据集涵盖英语和中文，包含单功能和并行功能案例。数据集分为多个子集，每个子集有不同的语言、音频类型、功能模式和大小。数据格式包括用户的口语查询、音频文件名、功能调用信息、时间信息、真实答案等。

SpeechFCEval is a collection of speech-based function-calling datasets built from existing open-source text datasets and constructed smart interaction scenarios. It covers English and Chinese, with single-function and parallel-function cases. The dataset is divided into several subsets, each with different languages, audio types, function modes, and sizes. The data format includes the users spoken query, audio file name, function call information, time information, ground truth, etc.