Prodorshok I

Prodorshok I是一个专为孟加拉语设计的孤立词数据集，由BRAC大学计算机科学与工程系创建。该数据集包含35位本地发言者在达卡录制的30个孟加拉语单词，总计1050个语音样本。数据集特别构建用于实现无手操作选择和数字界面导航的系统，包括10个数字词、10个方向词和10个位置词。创建过程涉及多阶段的语音增强，包括噪音减少和音频样本的合成增强。该数据集主要应用于自动语音识别（ASR）技术，旨在帮助视觉障碍人士通过语音命令轻松导航数字界面，增强人机交互体验。Prodorshok I is an isolated-word dataset specifically designed for the Bengali language, created by the Department of Computer Science and Engineering at BRAC University. It contains a total of 1050 speech samples of 30 Bengali words recorded by 35 local speakers in Dhaka. This dataset was specifically constructed for systems enabling hands-free selection and digital interface navigation, including 10 numerical words, 10 directional words, and 10 positional words. The dataset creation process involves multi-stage speech enhancement, including noise reduction and synthetic augmentation of audio samples. It is primarily applied to automatic speech recognition (ASR) technologies, aiming to help visually impaired individuals easily navigate digital interfaces via voice commands and enhance human-computer interaction experiences.