wav2gloss: Speech to Glossed Text of Low Resource Languages

• Guided by Prof. Lori, Prof. Shinji Watanabe and Prof. David Mortensen
• Automatically generate Interlinear Glossed Text of low-resource languages from audio files.
• Focusing on the speech enhancement of audio files using ESPNet2 to improve the ASR performance.

Multilingual TTS Accent Impressions for Accented ASR

• Guided by Prof. Shinji Watanabe and Prof. David Mortensen
• Fine-tuning of English ASR systems for L2-English speakers.
• Paper accepted to Text, Speech and Dialogue Conference (Sept 4-7, 2023)

Speaker Diariazation using Deep Learning Techniques

• Guided by Dr. Rajeev Rajan
• Implemented Bi-LSTM and Bi-GRU models with self-attention as baseline model and obtained an error rate of 25.7%.
• Experimented with various deep learning techniques involving transformers to diarize a given audio with data obtained from AMI Speech Corpus
• Paper accepted to ICNGIS 2022

Octoechos classification in liturgical music using Deep Learning frameworks

• Guided by Dr. Rajeev Rajan
• This automatic classification scheme is addressed using MTF-GRU, MTF-LSTM, MTF-SBU-LSTM, MTF-SB-GRU and MTF-SBU-GRU. The performance of these methods is compared using their accuracy (precision, recall and f1-score).
• The proposed experiment demonstrates the potential of SBU-LSTM in learning information through MTF.
• Paper accepted to INTERSPEECH 2022.

Understanding the scope of quantum technology to gauge its social impact - Independent and Self-started

• Presented a research poster for “Careers in Quantum” conducted by QET Labs and University of Bristol.
• Gave a talk on Quantum Technology and its Impact on Society for WomenTech Global Conference 2021.
• Gave a talk on Basics of Quantum Computing” as the first speaker of “Vocalize”, conducted by IEEE Travancore Hub on 4th October, 2020.