Harish Mallidi Machine Learning Scientist Alexa Science Amazon, Seattle email: last_name@jhu.edu www: http://mallidi.github.io

Interests

• Automatic speech recognition, Speech signal processing, Language identification, Speaker recognition, Machine learning.

Education

• June 2004 - August 2010, Bachelor of Technology + Master of Science (Electronics and Communication Engineering), https://www.iiit.ac.in/, India

• September 2011 - November 2016, Doctor of Philosophy, The Center for Language and Speech Processing (CLSP), Johns Hopkins University (JHU), USA, supervised by Prof. Hynek Hermansky
  
  
  • Thesis Proposal - Performance Monitor Techniques for Noise Robust Speech Recognition, December 2014.
  • Thesis - A Practical and Efficient Multistream Framework for Noise Robust Speech Recognition, February 2018.

Experience

• Participated as part of Brno University of Technology team in NIST LRE15 evals.

• Application of DNNs to classify languages directly, instead of conventional usage as bottleneck feature extractor.

• Participated as part of BBN team. (press release)

• Noise robust ASR based on FDLP based modulation features.

• Part of the team focusing on Unsupervised Confidence Estimation of Neural Networks

• Application: Noise Robust Speech Recognition

• Involved in implementation of 2D Convolutional Neural Networks (CNNs) in KALDI.

• Involved in the implementation of Neural Network based features in BBN's Byblos speech recognition toolkit.

• The features have been successfully used in 2014 DARPA RATS and IARPA BABEL evaluations.

Publications

• AM Castro Martinez, Sri Harish Mallidi, and B. T. Meyer, ``On the relevance of auditory-based Gabor features for deep learning in robust speech recognition'', Computer Speech & Language, Sept. 2017 (pdf)

• S. Ganapathy, Sri Harish Mallidi, and H. Hermansky, ``Robust Feature Extraction Using Modulation Filtering of Autoregressive Models'', IEEE Transactions on Audio, Speech and Language Proc. June 2014 (pdf)

• S. Garimella, Sri Harish Mallidi, and H. Hermansky, ``Regularized Auto-Associative Neural Networks for Speaker Verification'', IEEE Signal Processing Letters. Dec. 2012 (pdf)

• Sri Harish Mallidi, R. Maas, K. Goehner, A. Rastrow, S. Matsoukas, B. Hoffmeister ``Device-directed utterance detection'', Interspeech 2018. (pdf) (arXiv)

• Torres-Carrasquillo et. al. ``The MIT-LL, JHU and LRDE NIST 2016 speaker recognition evaluation system'', Interspeech 2017. (pdf)

• B. Meyer, Sri Harish Mallidi, H. Kayser, H. Hermansky ``Predicting error rates for unknown data in automatic speech recognition'', ICASSP 2017. (pdf)

• B. T. Meyer, Sri Harish Mallidi, H. Kayser, H. Hermansky ``Performance monitoring for automatic speech recognition in noisy multi-channel environments'', IEEE SLT 2016. (pdf)

• T. Ogawa, Sri Harish Mallidi, E. Dupoux, J. Cohen, N. Feldman, H. Hermansky ``A new efficient measure for accuracy prediction and its application to multistream-based unsupervised adaptation'', ICPR 2016. (pdf)

• Sri Harish Mallidi, H Hermansky ``A framework for Practical Multistream ASR'', Interspeech 2016. (pdf)

• Ruizhi Li, Sri Harish Mallidi, Lukas Burget, Oldich Plchot, Najim Dehak ``Exploiting Hidden-Layer Responses of Deep Neural Networks for Language Recognition'', Interspeech 2016. (pdf)

• Oldrich Plchot et. al.``BAT System Description for NIST LRE 2015'', Accepted in ISCA Speaker Odyssey, 2016.

• Sri Harish Mallidi, H Hermansky ``Novel Neural Network Based Fusion for Multistream ASR'', ICASSP 2016. (pdf) (poster)

• Sri Harish Mallidi, T Ogawa, H Hermansky ``Uncertainty Estimation of DNN classifiers'', IEEE ASRU 2015. (pdf) (poster)

• Sri Harish Mallidi, T Ogawa, K Vesely, P S Nidadavolu, H Hermansky ``Autoencoder based multi-stream combination for noise robust speech recognition'', Interspeech 2015. (pdf)

• Hynek Hermansky, et. al. ``Towards machines that know when they do not know: Summary of work done at 2014 Frederick Jelinek Memorial Workshop'', ICASSP 2015. (pdf)

• Tim Ng, et. al, ``Progress in the BBN Keyword Search System for the DARPA RATS Program'', Interspeech 2014. (pdf)

• P Matejka, L Zhang, T Ng, Sri Harish Mallidi, et. al. ``Neural Network Bottleneck Features for Language Identification'', ISCA Speaker Odyssey 2014. (pdf)

• Sri Harish Mallidi, Sriram Ganapathy and Hynek Hermansky, ``Robust Speaker Recognition Using Spectro-Temporal Autoregressive Models'', Interspeech 2013. (pdf)

• Jeff Ma, Bing Zhang, Spyros Matsoukas, Sri Harish Mallidi, Feipeng Li, Hynek Hermansky,``Improvements in Language Identification on the RATS Noisy Speech Corpus'', Interspeech 2013. (pdf)

• Pascal Clark, Sri Harish Mallidi, Aren Jansen, and Hynek Hermansky, ``Frequency Offset Correction in Speech without Detecting Pitch'', ICASSP 2013. (pdf)

• Oldrich Plchot, et.al, ``Developing a Speaker Identification System for the DARPA RATS Project'', ICASSP 2013. (pdf)

• Samuel Thomas, Sri Harish Mallidi, Thomas Janu, Hynek Hermansky, Nima Mesgarani, Xinhui Zhou, Shihab Shamma, Tim Ng, Bing Zhang, Long Nguyen and Spyros Matsoukas, ``Acoustic and Data-driven Features for Robust Speech Activity Detection'', Interspeech 2012. (pdf)

• Feipeng Li, Sri Harish Mallidi and H. Hermansky, ``Phone recognition in critical bands using sub-band temporal modulations'', Interspeech, 2012. (pdf) (poster)

• Samuel Thomas, Sri Harish Mallidi, Sriram Ganapathy and Hynek Hermansky, ``Adaptation Transforms of Auto-Associative Neural Networks as Features for Speaker Verification'', Odyssey Speaker and Language Recognition Workshop, Singapore, June 2012 (pdf)

• Daniel Garcia-Romero, Xinhui Zhou et al. The UMD-JHU 2011 Speaker Recognition System, ICASSP, Kyoto, Japan, March 2012 (pdf)

• Sri Harish Mallidi, Sriram Ganapathy and Hynek Hermansky, ``Modulation spectrum analysis for recognition of reverberant speech'', Interspeech, 2011. (pdf)

• Sri Harish Mallidi, Kishore S. Prahallad, Suryakanth V Gangashetty and B. Yegnanarayana, ``Significance of Pitch Synchronous Analysis for Speaker Recognition using AANN Models'', Interspeech, 2010. (pdf)

• Anand Joseph M., Sri Harish Mallidi and B. Yegnanarayana, ``Speaker Dependent Mapping of Source and System features for Enhancement of Throat Microphone Speech'', Interspeech, 2010. (pdf)

• Sudheer Kovela, Sri Harish Mallidi, Sri Rama Murty K and B. Yegnanarayana, ``Analysis of laugh signals for detecting in continuous speech'', Interspeech, 2009. (pdf)

Teaching

• Teaching Assistant for Processing of Audio and Visual Signals, Fall-2012, 2013, Spring-2016

• Teaching Assistant for Speech and Auditory Processing by Humans and Machines, Spring-2013, 2013