Subhadeep Dey

Short Biography

I am a PhD student at Ecole Polytechnic Federale de Lausanne (EPFL) in Electrical Engineering (EDEE) and research assistant at Idiap Research Institute under Dr. Petr Motlicek. I am funded by SIIP project which aims to build speaker identification system based on voice samples, meta-data and social media

My research focuses on building speaker identification system from voice samples in practical scenarios. I am interested in building system for short utterance and noisy conditions.

Research Interests

Speaker Verification, Speech Recognition, Machine Learning

Education

2013: Master of Science (by Research) from Computer Science Department, Indian Institute of Technology Madras.

Professional Experience

  • 2007 July - 2008 June: Project Assistant in Electrical Engineering, Indian Institute of Science, Bangalore.
  • 2012 July - 2013 October: Project Assistant in Computer Science Department, Indian Institute of Technology Madras.
  • 2013 April - 2014 April: Associate Project Engineer in Electrical and Electronics Department, Indian Institute of Technology Guwahati.
  • 2017 July - 2017 October: Internship in NEC Corporation, Japan.

Publications

  • S. Dey, T. Koshinaka, P. Motlicek and S. Madikeri, "DNN based speaker embedding using content information for text-dependent speaker verification", to appear in ICASSP 2018.
  • S.Dey, S. Madikeri, P. Motlicek and M. Ferras, "Content normalization for text-dependent speaker verification", in InterSpeech 2017, Stockholm.
  • S.Dey, P. Motlicek, S. Madikeri and M. Ferras, "Template matching for text-dependent speaker verification", in Speech Communication.
  • S. Dey, P. Motlicek, S. Madikeri and M. Ferras, "Exploiting sequence information for text-dependent speaker verification", in ICASSP 2017, New Orleans, USA
  • S. Madikeri, M. Ferras, P. Motlicek and S. Dey, "Intra-class covariance adaptation in PLDA back-ends for speaker verification", in ICASSP 2017, New Orleans, USA
  • M. Ferras, S. Madikeri, S. Dey , P. Motlicek and H. Bourlard, "Inter-task System Fusion for Speaker Recognition", in InterSpeech 2016, San Francisco, USA.
  • S.Dey, S. Madikeri, M. Ferras and P. Motlicek, "Deep Neural Networks Posterior for text-dependent speaker verification" in ICASSP 2016, Shanghai, China
  • S.Dey, S. Madikeri, and P. Motlicek, "Information Bottleneck Clustering for domain adaptation" in ICASSP 2016, Shanghai, China
  • M. Ferras, S. Madikeri, P. Motlicek, S.Dey and H. Bourlard,"A large-scale open-source acoustic simulator for speaker recognition", in IEEE Signal Processing Letters, 2016.
  • P. Motlicek, S.Dey, S. Madikeri, and L. Burget, "Employment of Subspace Gaussian Mixture Models in speaker recognition", in Proc. of ICASSP 2015, Brisbane, Australia, pp. 4445-4449.

Achievement

  • IEEE Ganesh N Ramaswamy memorial student grant sponsored by IBM for the best paper in the speaker/language recognition area in ICASSP 2017.

Links



Tel: +41 27 721 7702
Office: 305-8
Contact