Projects

1. Speech and NLP:

• Agency: TCS Research & Innovation, Mumbai

• Spoofing detection: ASVspoof 2021 challenge

• Automatic and explainable oral/speech fluency assessment

• End-to-end automatic speech recognition

2. DeepRASD: Deep learning for Robust and Adaptive Speaker Detection

• Agency: Applied Physics Lab JHU, USA

• Improved speaker clustering for speaker diarization using Generative Adversarial Network

• Robust speaker verication from far-eld and noisy recordings

3. CARE: Center for Autism Research in Engineering

• Agency: NIH, NIMH and NSF (USA)

• Improved x-vector diarization performance for child-psychologist interaction (USC CARE) data

4. GARD: DARPA in Guaranteeing AI Robustness against Deception

• Agency: DARPA (USA)

• Adversarial attack detection and defence for audio in speaker verication systems