BanglaNum
收藏资源简介:
BanglaNum是一个专为孟加拉数字语音识别设计的数据集,由孟加拉国工程技术大学电气与电子工程系创建。该数据集包含2252条语音样本,涵盖数字0至9,来源于40名孟加拉语母语者的录音。数据收集在教室环境中进行,使用MATLAB软件录制,确保了数据的真实性和日常应用性。数据集的创建旨在训练人工神经网络,以实现语音驱动的数字输入系统,特别适用于自动应答系统和电子商务网站的语音助手。该数据集的应用领域主要集中在提高孟加拉语语音识别的准确性和效率,解决现有语音识别系统在孟加拉语处理上的不足。
BanglaNum is a dataset tailored specifically for Bengali digit speech recognition, developed by the Department of Electrical and Electronic Engineering of Bangladesh University of Engineering and Technology. It comprises 2252 speech samples covering digits 0 through 9, recorded by 40 native Bengali speakers. All data was collected in a classroom setting and recorded using MATLAB software, ensuring the authenticity and daily practicality of the dataset. The dataset was constructed to train artificial neural networks for speech-driven digital input systems, particularly suitable for automatic answering systems and voice assistants on e-commerce websites. Its main application areas focus on improving the accuracy and efficiency of Bengali speech recognition, addressing the shortcomings of existing speech recognition systems in handling Bengali language.




