I am a researcher at the Université de Lorraine in France. I am conducting my research with the MultiSpeech team of LORIA-INRIA laboratory in Nancy and the language processing team at LIA, Université d'Avignon. I am working on techniques for modelling context in large vocabulary continuous speech recognition, in order to improve recognition of proper names in diachronic audio documents, as part of the ContNomina project. I am exploring Bayesian probabilistic models and continuous word representations from neural networks to model topics and context in large vocabulary continuous speech recognition.
Before starting my PhD at LORIA, I was a research engineer with the Speech & NLP group at TCS Innovation Labs - Mumbai; where I worked on several projects including semi-automatic construction of frugal speech corpora for resource deficient (Indian) languages using speech data on the internet, English language pronunciation evaluation tool, speech speedometer for real time measurement of speaking rate, automatic transcription of call centre conversations for analytics, speaker identification and verification tasks, speaker segmentation in speech, Indian language transliteration and speech enabled telephone IVR systems.
I received a Bachelors degree in Electronics Engineering from University of Mumbai in 2007. During my undergraduate studies I was also an IEEE student member and conducted workshops on building remote & computer controlled cars & hovercrafts and software simulation of analog, digital and communication circuits.