Contact | Education | Research and publications | Thesis | Professional | Teaching | Other skills | PDFPDF 

Research and publications

Publications

  1. Neural-FST Class Language Model for End-to-End Speech Recognition
  2. Improved pronunciation prediction accuracy using morphology
  3. Improved Neural Language Model Fusion for Streaming Recurrent Neural Network Transducer
  4. Anti-aliasing regularization in stacking layers
  5. Algorithmic Exploration of American English Dialects
  6. A Streaming On-Device End-To-End Model Surpassing Server-Side Conventional Model Quality and Latency
  7. Phoneme-Based Contextualization for Cross-Lingual Speech Recognition in End-to-End Models
  8. Better Morphology Prediction for Better Speech Systems
  9. Model Unit Exploration for Sequence-to-Sequence Speech Recognition
  10. Phoebe: Pronunciation-aware Contextualization for End-to-end Speech Recognition
  11. Lingvo: a Modular and Scalable Framework for Sequence-to-Sequence Modeling
  12. Dictionary Augmented Sequence-to-Sequence Neural Network for Grapheme to Phoneme prediction
  13. Sequence-to-Sequence Neural Network Model with 2D Attention for Learning Japanese Pitch Accents
  14. Pronunciation learning with RNN-transducers
  15. Learning Personalized Pronunciations for Contact Name Recognition
  16. NN-grams: Unifying neural network and n-gram language models for speech recognition
  17. On the compression of recurrent neural networks with an application to LVCSR acoustic modeling for embedded speech recognition
  18. Optimizing OPC data sampling based on orthogonal vector space
  19. Exploring the Nature of Trader Intuition
  20. Model-based scanner tuning in a manufacturing environment
  21. Investigating signal integration with canonical correlation analysis of fMRI brain activation data
  22. Human imagination in financial markets with insiders
  23. A Mind for the Market: an fMRI Study of Attribution of Mental States to Financial Markets
  24. SCR Recording During fMRI Acquisition

Patents

  1. Contextual biasing for speech recognition using grapheme and phoneme data
  2. Contextual biasing for speech recognition using grapheme and phoneme data
  3. Compressed recurrent neural network models
  4. Date and/or time resolution
  5. Learning personalized entity pronunciations
  6. Information matrix creation and calibration test pattern selection based on computational lithography model parameter
  7. Calibration pattern selection based on noise sensitivity
  8. Harmonic resist model for use in a lithographic apparatus and a device manufacturing method

Note: Additional US patent applications under review (closed to the general public).