Premove ITN is an open-source, context-aware inverse text normalizer built for voice agents. It turns messy ASR transcripts into structured values APIs can actually use: phone numbers, dates, times, amounts, IDs, emails, URLs, and more. Instead of generating normalized text directly, it deterministically produces valid candidates, uses a neural model to score them in sentence context, and applies exact decoding to choose compatible outputs. Open weights, Python package, real-time inference.

Hey everyone 👋 I started building Premove ITN while working on a voice agent. The problem looked simple at first: turn spoken forms like “one hundred twenty three” into 123. But the harder cases were contextual: meet me at two thirty → 2:30 room number two thirty → 230 I first tried deterministic rules, then NVIDIA Thutmose, then a BIO-tagging approach. What finally worked was separating the problem into three parts: 1. Deterministic code generates structurally valid candidates 2. A contextual neural model scores them 3. Exact decoding chooses a compatible final path That became Premove ITN. On a frozen 1,500-row synthetic benchmark, Premove reached 89.7% overall semantic accuracy, compared with 59.4% for NVIDIA Thutmose and 55.8% for text-processing-rs. On the dedicated voice-agent subset, Premove reached 99.5%. The code and model weights are open source. I’d especially love feedback from people building voice agents: what normalization failures are you still seeing in production?

Interested to see how the semantic accuracy holds up, great concept for an issue that is specific and tough to solve.

@david_cavill1 Hey David, thanks for checking it out!

Yes, I agree with you. Generic inverse text normalization is extremely hard to solve, but since I'm targeting the voice agent domain specifically, I've seen quite some progress.

Hope it helps people building voice agents.

Interested to see how the semantic accuracy holds up, great concept for an issue that is specific and tough to solve.

@david_cavill1 Hey David, thanks for checking it out!

Yes, I agree with you. Generic inverse text normalization is extremely hard to solve, but since I'm targeting the voice agent domain specifically, I've seen quite some progress.

Hope it helps people building voice agents.

Premove ITN

Open-source context-aware ITN for voice agents

Open-source context-aware ITN for voice agents