English audio-video to Marathi audio-video using Machine learning

Author(s)	Ms. Karishma Karande, Ms. Ankita Thepale, Ms. Vidhi Karade, Mr. Swapnil Kohle, Mr. Soham Wankhede, Mr. Mandar Deo, Mr. Chaitanya Thapa
Country	India
Abstract	In this paper, we explore different techniques of overcoming the challenges of low-resource in Neural Machine Translation (NMT), specifically focusing on the case of English-Marathi NMT. This report details the objective, methodology, and system overview for developing an expert-level Speech-to-Speech Translation (S2ST) system for the resource-constrained English-to-Marathi language pair using machine learning. The traditional cascaded approach (Automatic Speech Recognition -> Machine Translation -> Text-to-Speech) is critically assessed and deemed suboptimal due to its inherent susceptibility to compounded errors, high computational latency, and significant loss of prosodic information during the intermediate text representation stage. To circumvent these limitations, a Unit-to-Unit Sequence-to-Sequence (Seq2Seq) framework is proposed.
Keywords	Neural Machine Translation (NMT), Audio-Visual Machine Translation, English-to-Marathi Translation, Low-Resource Language Translation, Automatic Speech Recognition (ASR), Text-to-Speech (TTS), End-to-End Translation, Unit-to-Unit, Sequence-to-Sequence (Seq2Seq), Lip Synchronization (Lip-Sync).
Field	Engineering
Published In	Volume 7, Issue 6, November-December 2025
Published On	2025-11-16
DOI	https://doi.org/10.36948/ijfmr.2025.v07i06.60826

About IJFMR Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us		+91-9687-828-838	editor@ijfmr.com

International Journal For Multidisciplinary Research