NARD

NARD数据集是由法国国家信息与自动化研究所格勒诺布尔-阿尔卑斯分部创建，包含852条记录，旨在为家用机器人提供声音表示和分类的基准。该数据集在真实环境下使用NAO机器人录制，包含多种声音类别，如厨房、办公室、非言语和语音声音。数据集的创建过程涉及在不同场景下录制声音，并手动进行声音分割。NARD数据集主要应用于家用机器人的声音识别和分类，以提高机器人在复杂声学环境中的认知和交互能力。The NARD dataset was developed by the Grenoble-Alpes Branch of the French National Institute for Informatics and Automation (INRIA), containing 852 records. It is designed as a benchmark for voice representation and classification for home robots. This dataset was recorded in real-world environments using the NAO robot, and covers multiple sound categories including kitchen-related, office-related, non-verbal, and speech sounds. The dataset creation process included recording sounds in various scenarios and performing manual sound segmentation. The NARD dataset is primarily applied to voice recognition and classification for home robots, to enhance the robots' cognitive and interactive capabilities in complex acoustic environments.