Designed experiments, held office hours, code reviews, evaluate assignments, conducted lab sessions[EN.520.612] Machine Learning in Signal Processing[EN.520.659]Machine Learning in Medical Applications [EN 520.123] Computational Modeling for Electrical and Computer Engineering

Designed and taught course "Hands-on Speech Technologies and Applications with Python" to 10 undergraduate students.

Robust Speech and Speaker Recognition, Speech and NLP Group↪Projects:✶ Robust Speech Recognition• Investigated the performance of Time Delay Neural Network - Denoising Auto-encoder (TDNN-DAE) in seen and unseen noise conditions and identified the problem of inferior performance in unseen conditions.• Proposed an enhanced TDNN-DAE using two front-end enhancement techniques, Spectral Subtraction (SS) and Vector Taylor Series - Auditory Model (VTS-AM) • Developed pseudo-SNR based system switching to maintain the performance across seen and unseen noise conditions. ✶ CHiME-5 Challenge• Collaborated with Indian Institute of Technology Bombay to solve the CHiME-5 challenge which addressed the problem of distant multi-microphone conversational speech recognition in everyday home environments.• Proposed a solution to enhance the beamformed utterances by using a TDNN-DAE to make them acoustically similar to the close-talking microphone utterances. ✶ mKRISHI Project• The mKRISHI project harnesses Information Technology to cater to the needs of the farmers in the rural sector.• Developed an automatic speech recognition system using Kaldi SRILM toolkit for Hindi language to identify the topic of the inquiry phone calls and route them to domain experts without manual intervention. ✶ Language Modelling for Speech Enabled Chatbot• Worked with a team to build a cloud-based speech-enabled chatbot to answer queries on company policies.• Built and optimized an in-domain language model using Kaldi. ✶ WhoAmI - a Speaker Verification demo• Implemented a speaker verification proof of concept using Voice Biometry Standard Toolkit. • Released the first version along with webpage and step-by-step user manual.• Led knowledge transfer by providing clear and thorough explanations by walking through the codebase.

Teaching assistant for the the following courses:- [EE322] Digital Signal Processing : Designed and graded problem sets and helped keeping student logs, invigilated examinations- [EE312] Microprocessors and Microcontrollers : Reinforced lessons taught in classroom by taking tutorials for practicals- [EE619] Sensors in Instrumentation : Kept student logs for grades and attendance- [EE121] Basic Electronics Engineering : Graded assignments and kept student logs for grades and attendance