HANNA Voice Commands Dataset

This dataset consists of recordings for device provisioning and configuration recorded from 10 different speakers, 5 female, 5 male with different levels of English language proficiency. The recording are also mixed with noise corresponding to factory, office, talking and traffic backgrounds to emulate corresponding environments. It was used to evaluate the performance of various speech to text engines.

本数据集收录了由10名不同受试说话人录制的设备预配置与配置相关录音，其中包含5名女性、5名男性，受试对象的英语熟练度水平各异。为模拟真实应用环境，所有录音均混入了工厂、办公室、日常交谈及交通场景的背景噪声。该数据集曾被用于评估各类语音转文字（Speech-to-Text）引擎的性能表现。