US Patents: Robust Speech Recognition

Foundational robustness work, patented.

Two granted US patents from my work on robust speech recognition:

  • US 11,335,329: Method and System for Generating Synthetic Multi-Conditioned Data Sets for Robust ASR. Filed Mar 2020, granted 2022.
  • US 11,322,156: Features Search and Selection Techniques for Speaker and Speech Recognition. Filed Dec 2019, granted 2022.

This synthetic-data and feature-robustness work predates, and anticipates, much of today’s interest in data augmentation for reliable speech models.