This inventor holds 1 USPTO granted patent and 4 published patent applications. Top assignee: Google Inc.. Active years: 2026.
Company Filing History:
Years Active: 2026
Title: Innovations of Kyle Scott Kastner
Introduction
Kyle Scott Kastner is an accomplished inventor based in Waltham, MA (US). He has made significant contributions to the field of automatic speech recognition. His innovative approach combines audio and textual representations to enhance the accuracy of speech recognition systems.
Latest Patents
Kastner holds a patent for a method titled "Automatic speech recognition accuracy with multimodal embeddings search." This method involves receiving training data that includes a set of transcribed speech utterances, each paired with a corresponding transcription. For each utterance, the method generates an encoded audio representation and an encoded textual representation. It also creates higher-order audio and textual feature representations, determining a loss based on these features. The method trains a speech encoder and a text encoder of a correction model based on the determined loss for each transcribed speech utterance.
Career Highlights
Kyle Scott Kastner is currently employed at Google Inc., where he continues to work on advancements in speech recognition technology. His expertise in multimodal embeddings has positioned him as a valuable asset in the tech industry.
Collaborations
Kastner collaborates with fellow innovator Christopher Li, contributing to the development of cutting-edge technologies in their field.
Conclusion
Kyle Scott Kastner's work in automatic speech recognition showcases his innovative spirit and dedication to improving technology. His contributions are paving the way for more accurate and efficient speech recognition systems.