Fact · Speech recognition
Google said its 2019 on-device Gboard speech recognizer uses a model trained with RNN transducer technology that is compact enough to reside on a phone.
Checked against its source on October 17, 2026
In the source's own words
“we present a model trained using RNN transducer (RNN-T) technology that is compact enough to reside on a phone”
An All-Neural On-Device Speech Recognizer, Johan Schalkwyk, Google Research Blog, 2019, Introduction of the post.
Open the original- What it measures
- This is the production use of a transducer for dictation on a phone, described by Google's speech team in March 2019, and it matches the paper by He and colleagues.
- What it does not show
- A 2019 product description. Later releases may use newer models, and the post is about one recognizer for one language at launch.