Fact · Speech recognition
Google says its RNN-T recognizer continuously processes input audio and streams output symbols, a property that suits speech dictation, and in its implementation the symbols are characters.
Checked against its source on October 17, 2026
In the source's own words
“continuously processes input samples and streams output symbols, a property that is welcome for speech dictation. In our implementation, the output symbols are the characters of the alphabet.”
An All-Neural On-Device Speech Recognizer, Johan Schalkwyk, Google Research Blog, 2019, Body of the 2019 Google Research blog post.
Open the original- What it measures
- The post describes the model that runs on the phone for Gboard voice typing. Streaming means the text can appear while you speak.
- What it does not show
- A company blog post describing one product. It does not report word error rates in this passage.