Ravichander Vipperla

Post Doctoral Researcher, Multimedia Communications, Eurecom, France

2229 route des cretes, BP 193, F-06560 Sophia-Antipolis
(+33) (0)4 93 00 82 62
(+33) (0)4 93 00 82 00
firstname.lastname@eurecom.fr

Research

I am a researcher in speech technology and primarily focus on the machine listening aspect of it. I am interested in understanding how humans perceive audio signals and in replicating such functionality on computing devices. My research efforts include the following topics: automatic speech recognition (ASR), speaker adaptation, ASR for ageing voices, speaker recognition and diarization, audio source separation and signal enhancement.

Here at Eurecom, I am associated with the following two projects:

ALIAS

 The primary objective of the Adaptable Ambient LIving ASsistant (ALIAS) project is the product development of a mobile robot system that interacts with elderly users, monitors and provides cognitive assistance in daily life, and promotes social inclusion by creating connections to people and events in the wider world. My work comprises development and integration of software modules for user identification based on voice.

TABULA RASA

 The Tabula-Rasa project aims to address the vulnerability of a range of biometric authentication systems to spoofing attacks and to propose robust anti-spoofing countermeasures. I look at the problem of attacking speaker verification systems using synthetic speech generated with parametric models.

Education

Ph.D., Informatics (2011)
University of Edinburgh, UK
Thesis: Automatic speech recognition for ageing voices. (pdf)
Advisors: Prof. Steve Renals and Dr. Joe Frankel
Awarded PhD studentship by the Scottish funding council under the MATCH project.
M.Tech., Information and Communication Technology (2004)
DA-IICT, India
Received the president's gold medal for excellence in academics
B.Engg., Electronics and telecommunications (2002)
University of Mumbai, India

Past work experience

Engineer, Emerging Technologies and services (Apr 2004 - Aug 2006)
Reliance Communications, India
In a small team, we set out and achieved an ambitious project of building a fully functional prototype of value added services powered by voice based interface for mobile telephone users. My contributions to the project include active role in speech corpus collection in 10 Indian languages, training and adapting acoustic models, building the lexicon for these languages, performance evaluation, development of tools to monitor speech recognition accuracies and dialogue success rates and in the design of voicexml applications. It was great fun!

Publications

Journal
  • Direct posterior confidence for out-of-vocabulary spoken term detection. Dong Wang, Simon King, Joe Frankel, Ravichander Vipperla, Nicholas Evans, Raphael Troncy. (to appear) ACM Transactions on Information systems, 2012
  • Evolutionary discriminative confidence estimation for spoken term detection. Javier Tejedor, Alejandro Echeverria, Dong Wang, and Ravichander Vipperla. Multimedia tools and applications, Springer, 2011
  • Ageing voices: the effect of changes in voice parameters on ASR performance. Ravichander Vipperla, Steve Renals, and Joe Frankel. EURASIP Journal on Audio, Speech, and Music Processing, 2010
Conference
  • Phone Adaptive Training for Speaker Diarization. Simon Bozonnet, Ravichander Vipperla and Nicholas Evans. Interspeech 2012, Portland (Oregon, USA)
  • Convolutive Non-Negative Sparse Coding and New Features for Speech Overlap Handling in Speaker Diarization. Juergen Geiger, Ravichander Vipperla, Simon Bozonnet, Nicholas Evans, Bjoern Schuller and Gerhard Rigoll. Interspeech 2012, Portland (Oregon, USA)
  • Spoofing countermeasures for the protection of automatic speaker recognition systems against attacks with artificial signals. Federico Alegre, Ravichander Vipperla and Nicholas Evans. Interspeech 2012, Portland (Oregon, USA)
  • Speech Overlap Detection Using Convolutive Non-Negative Sparse Coding: New Improvements and Insights. Juergen Geiger, Ravichander Vipperla, Nicholas Evans, Bjoern Schuller and Gerhard Rigoll. Eusipco, 2012, Bucharest (Romania)
  • On the Vulnerability of Automatic Speaker Recognition to Spoofing Attacks with Artificial Signals. Federico Alegre, Ravichander Vipperla, Nicholas Evans and Benoit Fauve. Eusipco, 2012, Bucharest (Romania)
  • Speech overlap detection and attribution using convolutive non-negative sparse coding. Ravichander Vipperla, Juergen Geiger, Simon Bozonnet, Dong Wang, Nicholas Evans, Bjoern Schuller and Gerhard Rigoll. IEEE International Conference on Acoustics Speech and Signal processing, 2012, Kyoto (Japan)
  • Robust speech recognition in multi-source noise environments using convolutive non-negative matrix factorization. Ravichander Vipperla, Simon Bozonnet, Dong Wang and Nicholas Evans. First International Workshop on Machine Listening in Multisource Environments, Interspeech, 2011, Florence (Italy)
  • Online pattern learning for non-negative convolutive sparse coding. Dong Wang, Ravichander Vipperla and Nicholas Evans. Interspeech 2011, Florence (Italy)
  • Parallel and hierarchical decision making for sparse coding in speech recognition. Dong Wang, Ravichander Vipperla and Nicholas Evans. Interspeech 2011, Florence (Italy)
  • Augmentation of adaptation data. Ravichander Vipperla, Steve Renals and Joe Frankel. Interspeech 2010, Makuhari (Japan)
  • Age recognition for spoken dialogue systems: Do we need it? Maria Wolters, Ravichander Vipperla and Steve Renals. Interspeech 2009, Brighton (UK)
  • Speech Input from Older Users in Smart Environments: Challenges and Perspectives. RaviChander Vipperla, Maria Wolters, Kallirroi Georgila, and Steve Renals. HCI International 2009, San Diego (USA)
  • Longitudinal study of ASR performance on ageing voices. Ravichander Vipperla,Steve Renals and Joe Frankel. Interspeech 2008, Brisbane (Australia)
Technical reports
  • Online non-negative convolutive pattern learning for speech signals. Dong Wang, Ravichander Vipperla, Nicholas Evans, Thomas Fang Zheng. Research report - RR-11-261, Eurecom, France
  • Speech overlap detection using convolutive non-negative sparse coding. Ravichander Vipperla, Dong Wang, Simon Bozonnet, Nicholas Evans. Research report - RR-11-257, Eurecom, France