Jeffrey John Farris

Title: Innovations of Jeffrey John Farris

Introduction

Jeffrey John Farris is an accomplished inventor based in Crystal Lake, IL (US). He has made significant contributions to the field of automatic speech recognition, holding 2 patents that showcase his innovative approaches to improving speech-to-text technologies.

Latest Patents

One of his latest patents is titled "Removing bias from automatic speech recognition models using internal language model estimates." This invention addresses the challenge of bias in automatic speech recognition model predictions. By processing audio data through an automatic speech recognition model, Farris's method allows for the comparison of original word token predictions with masked audio predictions. This comparison provides an estimate of an internal language model, which can be used to modify predictions and remove lexical bias, ultimately enhancing speech prediction accuracy.

Another notable patent is "Personalized batch and streaming speech-to-text transcription of audio." This invention focuses on reducing the error rate of automatic speech recognition systems when transcribing rare and out-of-vocabulary words. Farris's techniques achieve personalization of connectionist temporal classification models through adaptive boosting, which biases sub-word predictions towards rare long-tail words. This innovation improves the accuracy of speech-to-text transcription without requiring the ASR system to be trained on a custom vocabulary.

Career Highlights

Jeffrey John Farris is currently employed at Amazon Technologies, Inc., where he continues to develop cutting-edge technologies in speech recognition. His work has significantly impacted the efficiency and accuracy of automatic speech recognition systems.

Collaborations

Farris has collaborated with notable colleagues, including Monica Lakshmi Sunkara and Srikanth Ronanki, contributing to advancements in their shared field of expertise.

Conclusion

Jeffrey John Farris is a pioneering inventor whose work in automatic speech recognition has led to innovative solutions that enhance the accuracy of speech-to-text technologies. His contributions continue to shape the future of this important field.