Arman-AV

Arman-AV数据集是由伊朗科学与技术大学计算机工程学院创建的多用途视听数据集，专门针对波斯语。该数据集包含约220小时的视频，涉及1760名不同的演讲者，适用于自动语音识别、视听语音识别和说话人识别等多种任务。数据集内容丰富，包括从访谈、电影和用户生成内容中收集的视频，旨在解决低资源语言中大型数据集缺乏的问题。创建过程中，研究人员利用自动化的数据收集和处理流程，确保数据的质量和多样性。Arman-AV数据集的应用领域广泛，不仅限于语音识别，还包括唇读和说话人识别等，为波斯语的视听处理技术提供了重要的资源。

The Arman-AV dataset is a versatile audiovisual dataset developed by the School of Computer Engineering, Iran University of Science and Technology, specifically for the Persian language. It contains approximately 220 hours of video content involving 1760 distinct speakers, and supports multiple tasks including automatic speech recognition (ASR), audiovisual speech recognition, and speaker recognition. Featuring rich content, the dataset collects videos from interviews, films, and user-generated content, aiming to alleviate the shortage of large-scale datasets for low-resource languages. During its development, researchers utilized automated data collection and processing workflows to ensure the dataset's quality and diversity. The Arman-AV dataset has a wide range of application scenarios beyond speech recognition, covering lip reading and speaker recognition as well, serving as a critical resource for audiovisual processing technologies of the Persian language.