Skip to main content

​
Available now: Premove ITN

Speech recognition returns words. Applications often need written values: “four thirty” might be a time, while “one oh five” might be a room code. Premove ITN uses sentence context to choose a valid written form for English voice-agent transcripts. It is one available part of the broader Premove effort.

Explore ITN

See what Premove ITN does and where it fits.

Get started

Install the package and normalize your first transcript.

View results

Read the benchmark method, results, and limitations.
The main Premove project is still in development. More documentation will appear here as its work is ready to share.