Inventor · Sunnyvale, CA, US

Trevor Strohman

26Patents
4h-index
81Co-inventors
62Inventor score

Filing activity: Oct 26, 2010 → May 31, 2024

Most-cited inventions

PatentTitleAreaCited byStatus
US8489604B1 Automated resource selection process evaluation Physics 17 Active
US9741339B2 Data driven word pronunciation learning and scoring with crowd sourcing based on the word's phonemes pronunciation scores Physics 8 Active
US8959020B1 Discovery of problematic pronunciations for automatic speech recognition systems Physics 6 Active
US8892597B1 Selecting data collections to search based on the query Physics 5 Active
US11763813B2 Methods and systems for reducing latency in automated assistant interactions Physics 4 Active
US11594212B2 Attention-based joint acoustic and text on-device end-to-end model Physics 1 Active
US11715458B2 Efficient streaming non-recurrent on-device end-to-end model Physics 1 Active
US12027154B2 Emitting word timings with end-to-end models Physics 0 Active
US12354598B2 Rare word recognition with LM-aware MWER training Physics 0 Active
US12126845B2 Ephemeral learning of machine learning model(s) Electricity 0 Active
US12183322B2 Language agnostic multilingual end-to-end streaming on-device ASR system Physics 0 Active
US12417770B2 Unified cascaded encoder ASR model for dynamic model sizes Physics 0 Active
US12315497B2 Intended query detection using E2E modeling for continued conversation Physics 0 Active
US12361927B2 Emitting word timings with end-to-end models Physics 0 Active
US12051416B2 Methods and systems for reducing latency in automated assistant interactions Physics 0 Active
US12412566B2 Lookup-table recurrent language model Physics 0 Active
US12190869B2 Optimizing inference performance for conformer Physics 0 Active
US12340799B2 Identifying and correcting automatic speech recognition (ASR) misrecognitions in a decentralized manner Physics 0 Active
US12051404B2 Efficient streaming non-recurrent on-device end-to-end model Physics 0 Active
US11580956B2 Emitting word timings with end-to-end models Physics 0 Active
US12020703B2 Enabling natural conversations with soft endpointing for an automated assistant Physics 0 Active
US12354595B2 Deliberation by text-only and semi-supervised training Physics 0 Active
US12118988B2 Transducer-based streaming deliberation for cascaded encoders Physics 0 Active
US12354597B2 Disfluency detection models for natural conversational voice systems Physics 0 Active
US12406147B2 Dialog management for large language model- based (LLM-based) dialogs Physics 0 Active

Source: USPTO / EPO open patent data. Inventor disambiguation is heuristic; counts are objective bibliographic measures.