A Multi-Purpose Audio-Visual Corpus for Multi-Modal Persian Speech Recognition: the Arman-AV Dataset

TL;DR

This paper introduces Arman-AV, a comprehensive multi-modal Persian dataset with 220 hours of videos, enabling advancements in lip reading, speech, and speaker recognition for a low-resource language.

Contribution

It presents the first large-scale Persian lip reading dataset and a viseme detection technique that improves lip reading accuracy, supporting multiple audio-visual recognition tasks.

Findings

The dataset includes 220 hours of videos from 1760 speakers.

A viseme detection method increases lip reading accuracy by 7%.

Baseline models are provided for various recognition tasks.

Abstract

In recent years, significant progress has been made in automatic lip reading. But these methods require large-scale datasets that do not exist for many low-resource languages. In this paper, we have presented a new multipurpose audio-visual dataset for Persian. This dataset consists of almost 220 hours of videos with 1760 corresponding speakers. In addition to lip reading, the dataset is suitable for automatic speech recognition, audio-visual speech recognition, and speaker recognition. Also, it is the first large-scale lip reading dataset in Persian. A baseline method was provided for each mentioned task. In addition, we have proposed a technique to detect visemes (a visual equivalent of a phoneme) in Persian. The visemes obtained by this method increase the accuracy of the lip reading task by 7% relatively compared to the previously proposed visemes, which can be applied to other languages as well.