Text Normalization (ITN)
Convert spoken text to written form using Inverse Text Normalization (ITN). Essential for accurate Word Error Rate (WER) calculations and transcript comparison.
What is Text Normalization?
ITN transforms spoken-form text into written form. For example: "twenty three dollars" → "$23", "two thousand twenty six" → "2026". This is crucial for comparing transcripts with different formatting. Powered by NVIDIA NeMo for multilingual support.
Normalize Text
Enter text as it would be spoken aloud
Try These Examples
Use Cases
📊 WER Calculation
Normalize both reference and hypothesis transcripts before calculating Word Error Rate to avoid false errors from different number representations.
📝 Transcript Comparison
Compare transcripts from different providers by normalizing them to a standard format.
🔍 Search & Indexing
Convert spoken queries to written form for better search results in transcript databases.
Supported Languages
Currently supports 12+ languages including English, Russian, German, Spanish, French, Italian, Japanese, Chinese, and Korean. Powered by NVIDIA NeMo.