Hi I'm Rajat Hebbar. I am a PhD Student in the Signal Analysis and Interpretation Laboratory (SAIL) at USC.
My research work involves developing techniques for automatic processing of speech and audio signals in real-world settings.
In my time at SAIL, I have developed noise-robust machine learning models for preliminary speech-pipeline modules such as speech activity detection and gender identification, targeted to domains with challenging acoustic environments such as movies, and real-world egocentric audio.
PS: I also like to play chess.
Scripting languages
Packages/Tools
PS:
Disclaimer: The skill-levels depicted in this section are products of the author's self-adjudication. Any resemblance to actual skill levels (living or dead?) is purely coincidental
Think of these numbers as confidence scores
National Institute of Technology Karnataka
Major: Electronics and Communication Engineering
2012 - 2016
GPA: 8.85/10
University of Southern California
Major: Electrical Engineering (Signal Processing)
2016 - 2018
GPA: 3.85/4
University of Southern California
Major: Electrical Engineering
2018 - Present
GPA: 3.8/4
Large-scale self-supervised mining of 169K face-tracks from 240 movies, leveraging temporal/spatial co-occurrence of faces to mine positive/negative samples. Multiview adaptation of face-representations outperforms triplet learning for face-clustering on benchmark dataset.
Multiple-instance learning approach to detect foreground-speaker speech in egocentric audio recordings. Two-fold detection and localization of foreground segments using existing and novel pooling methods, transfer learned using SAD embeddings.
Flask-based web-application to deploy state-of-the-art neural network models for gender-based speaking time estimation from audio. Redis Queue (RQ) serves asynchronous background jobs for processing multiple files.
Automatic scalable method for extracting labeled data for speech activity detection (SAD) in movies, generating over 100 hours of aligned audio. Proposed lightweight CNN architectures to achieve state-of-the-art performance in movie-SAD, outperforming LSTM and ResNet models! Read more here!
Transfer learning of audio-event VGGish embeddings for gender identification. Trained neural-network models on weakly-labelled AudioSet data to outperform GMM based models in movies.
643 W 30th Street, Los Angeles, California 90007. (Please wait until post-COVID for Apt #)
+323 948 7643 (Please don't call me!)