Work

Architect, Samsung R&D Institute India April 2022 - Present 

Bengaluru, Karnataka

-> Exploring E2E ASR architectures for On Device ASR

-> Speech Recognition solutions for Samsung's Bixby Voice Intelligent Assistant 


Senior Research Scientist, Zapr Media Labs:  Jan 2020 - April 2022

Zapr Media Labs

Bengaluru, Karnataka

Developing end-to-end solution for Conversational AI

-> Developing Automatic speech recognition (ASR) for Indian Languages.

-> Exploring E2E ASR architectures like transformer, conformers and Unsupervised architectures like Wav2Vec, Hubert, WavLM

-> Context-based ASR and its application to Voice bot.

-> ASR with limited data using transfer learning approach.

-> ASR API level features like punctuation, confidence score, timing information etc.

-> Developing Neural Speech Synthesis for Indian Languages.

-> Voice cloning for any speaker with limited amount of data.

-> Developing Speech modules for Audit Analytics and Voice bot projects.

-> Managing research team to build solutions for conversational AI.

-> Building plans to acquire skills (hiring / training) and planning/integrating research road map with company strategy.


Machine Learning Scientist, Apple: Nov 2018 - Dec 2019 (1 year 2 months)

Apple Cambridge UK Division

Worked as contractor in Apple for developing Neural based TTS.

-> Developing Neural based Text to Speech Synthesis (TTS) in Noisy Scenario using Adaptation

-> Voice conversion and Speech enhancement using Generative Adversarial Network.


Postdoctoral Researcher, University of Crete: Jul 2017 - Dec 2019 (2 years 6 months )

University of Crete

Heraklion, Crete, Greece

Working on Project for Toshiba’s Cambridge Research Laboratory (2017-2019)

-> Children automatic Speech recognition using Prosody modifications, different acoustic features, data augmentation

-> Data augmentation using Voice conversion for Speech Recognition.

-> Voice conversion and Speech enhancement using Generative Adversarial Network.

-> Speech synthesis in noisy scenario using adaptation.

-> Exploring different acoustic features for WaveNet vocoder.

-> Taking guest lectures in Voice processing course in University of Crete


Research Associate, IISc: Feb 2017 - Jun 2017 (5 months)

Indian Institute of Science (IISc)

Bengaluru Area, India

-> Pitch estimation using Riesz transform.

-> Riesz transform for statistical parametric speech synthesis.

-> Exploring Riesz transform for Speech recognition application


Research Scholar, IIT Guwahati: Jan 2012 - Dec 2016 (5 years)

Indian Institute of Technology, Guwahati

Guwahati Area, India

-> Research and development of a Text to Speech System (TTS) in Assamese and Manipuri Languages.

-> Smaller TTS foot print is done using the Hidden Markov Model with HTS framework.

-> Improving the naturalness of statistical parametric speech synthesis using features derived from glottal activity region of speech.

-> Taken guest lectures for Speech processing course.

-> Main teaching assistant for Signals and Systems, and Speech processing course.


Software Engineer, Alcatel Lucent India Private Limited: Jul 2008 - Dec 2011 (3 years 6 months)

Alcatel-Lucent 

Banaglore

Working domain: NMS/EMS

Project: Optical Network management system

Details: Worked on research and development of next generation high leverage optical transport networks in DWDM, SONET, SDH and PKT technologies. Which includes the testing of optical management system over the optical network elements including optical transport service switches (TSS), photonic service switches (PSS), Light Managers (LM), Optical multiplexers, Ethernet interface units, TDM interface units.


Project Trainee, NAL : Jun 2007 - Jun 2008 (1 year 1 month)

National Aerospace Laboratories, Bangalore

Project: Flow separation and Control for Low Speed Wind Tunnel

Domain: Embedded design

Description: It is a implementing of hardware using PIC Microcontroller to detect the flow separation and control it using free jet

Hardware used: PIC16F877, Hitachi44780U LCD Display, Keypad(4x3) with MM74C922 decoder, Subtracter, Band pass filter, Level shifter and Amplifier, DC motor and Hot wire signal.

Software used: MPLABIDE v7.2, PIC C Compiler, IC Prog Flash Programmer, PIC Simulator