Skip to main content
Premove ITN is an open-source, context-aware inverse text normalization system for English voice-agent transcripts. It converts spoken ASR text into written values such as phone numbers, dates, times, money, email addresses, URLs, and identifiers. The v0.3.0 runtime adds one-inference structured results and deterministic temporal enrichment. Callers can request exact spans and resolved date values without triggering a second model pass.

​
Start here

Install Premove ITN and normalize your first transcript.

​
Integrate

​
Understand

​
Run in production

​
Go deeper

The implementation architecture, model provenance, and other internal documents explain how the release is built and verified. For product context, see the Premove ITN overview. For measured results, see the benchmark.