Pradeep R Applied Research Scientist, Swiggy Former TCS Research Scholar (PhD Student) Department of Computer Science and Engineering Indian Institute of Technology Kharagpur Address : Research Projects Lab, Takshashila Second Floor                   Department of Computer Science and Engineering,                   IIT Kharagpur, West Bengal 721 302, INDIA. Email:       pradeep [underscore] raj31 [at] iitkgp [dot] ac [dot] in

About

Research Interests

• Automatic Speech Recognition
• Decoding Graphs
• Lattice Structures
• Probabilistic Graphical Models
• Machine Learning

Education

• 2014-2019:   Ph. D in Speech Signal Processing from Indian Institute of Technology Kharagpur .
  PhD Thesis : Incorporation of Manner Of Articulation Knowledge to Improve the Performance of Automatic Speech Recognition System
  Supervisor : Prof. K Sreenivasa Rao


• 2012-2014:   M.Tech. in Signal Processing from Siddaganga Institute of Technology Tumkur .
  Masters Thesis : Exploiting Deep Neural Networks for Speech Recognition
  Advisor : Dr. R Kumaraswamy


• 2008-2012:   B.E. in Electronics and Communication Engineering from Visvesvaraya Technological University
  Bachelor Thesis : An assistive portable device and gesture control for speech and hearing impaired disabilities using body sensor network (BSN) technology.
  Advisor : Mr. Pradeep G


• 2008:   Higher Secondary (Pre-University Board)
Board:   Dept. of Pre-University Education, Karnataka


• 2006:   SSLC (Class X Board Exam)
Board:   Karnataka Secondary Education Examination Board

Publications      

Journals

1. Incorporation of Manner of Articulation Constraint in LSTM for Speech Recognition
   Pradeep R, K S Rao
   Circuits, Systems and Signal Processing (CSSP), pp. 1-19, 2019. [PDF].


2. Lattice Rescoring using Manner of Articulation Knowledge to Improve Speech Recognition Performance
   Pradeep R, K S Rao
   Speech Communication, 2018 (Revision Submitted).


3. Manner of Articulation Detection Constraint in the Decoding Graph for Improving Phoneme Recognition Accuracy
   Pradeep R, K S Rao
   Circuits, Systems and Signal Processing (CSSP), 2019 (Under Review).


4. Retraining End-to-End ASR System by Exploiting Manner of Articulation Detection Weights Derived using Connectionist Temporal Classification
   Pradeep R, K S Rao and Murari Akshai
   IEEE Transactions on Multimedia, 2019 (Under Review).

Conferences

1. Modifying LSTM Posteriors with Manner of Articulation Knowledge to Improve Speech Recognition Performance
   Pradeep R, K S Rao
   17^th IEEE International Conference on Machine Learning and Applications (ICMLA), Orlando, USA, December, 2018 [PDF].


2. Manner of Articulation Based Split Lattices for Phoneme Recognition
   Pradeep R, K S Rao
   24^th National Conference on Communications (NCC-2018) , IIT Hyderabad, India, February, 2018.[PDF]


3. Split Acoustic Modeling in Decoder for Phoneme Recognition
   Pradeep R, K S Rao
   14^thIEEE India Council International Conference (INDICON), IIT Roorkee, India, December, 2017.[PDF]


4. Deep neural networks for kannada phoneme recognition
   Pradeep R, K S Rao
   9^th International Conference on Contemporary Computing (IC3), Noida, India, August, 2016.[PDF]

Work Experience

• Senior Scientific Officer
Sponsor:   Department of Information Technology (DIT), New Delhi
Area : Speech-based Access of Agricultural Commodity Prices and Weather Information in Bihari Hindi Language
Duration:   October 2014 - October 2016

• TCS Research Scholar
Sponsor:   Tata Consultancy Services (Cycle-12)
Duration:   October 2016 - June 2019

• AI Researcher
Company:   Artivatic.ai
Area : Voice and Speech Research
Duration:   July 2019 - January 2020

Activities

• Advanced Topics in Speech Signal Processing
• Programming and Data Structures
• Computer Organization and Architectures