VAST: The Virtual Acoustic Space Traveler Dataset
收藏资源简介:
VAST数据集是由法国国家信息与自动化研究所雷恩-大西洋布列塔尼分部和印度理工学院坎普尔分校合作创建的,旨在通过虚拟声学空间旅行技术改善声音源定位。该数据集包含超过110,000个模拟房间脉冲响应(RIR),用于训练机器学习模型从音频特征映射到所需的音频属性。数据集设计考虑了多种室内环境,如办公室、大学和现代住宅,以确保其广泛的代表性。创建过程中使用了高效的C++/MATLAB房间声学模拟器,生成了各种声学环境的RIR。VAST数据集的应用领域包括声音源定位和双耳听觉,旨在解决传统基于到达时间差的声音定位方法的局限性。
The VAST dataset was collaboratively developed by the Rennes Bretagne-Atlantique branch of the French National Institute for Research in Computer Science and Automation (INRIA) and the Indian Institute of Technology Kanpur. It was created to enhance sound source localization via virtual acoustic space travel technologies. This dataset contains over 110,000 simulated room impulse responses (RIRs), which are utilized to train machine learning models for mapping audio features to target audio attributes. The dataset was designed to encompass a diverse range of indoor environments, including offices, university campuses, and modern residences, to ensure broad representativeness. An efficient C++/MATLAB room acoustics simulator was employed during the dataset development to generate RIRs for various acoustic environments. The application domains of the VAST dataset include sound source localization and binaural hearing, aiming to address the limitations of traditional sound localization methods based on time difference of arrival.

- 1VAST : The Virtual Acoustic Space Traveler Dataset法国国家信息与自动化研究所雷恩-大西洋布列塔尼分部 · 2016年



