Inverse text normalization (ITN)
Inverse text normalization converts spoken forms into written conventions — 'twenty twenty six' becomes '2026', 'ten thirty a m' becomes '10:30 AM', 'doctor smith' becomes 'Dr. Smith'.
ITN is why usable transcripts do not spell out every number and address. It is locale-dependent (dates, currencies and phone formats differ by country) and one of the quiet quality gaps between transcription systems.
Related terms
Frequently asked
What is Inverse text normalization?
Inverse text normalization converts spoken forms into written conventions — 'twenty twenty six' becomes '2026', 'ten thirty a m' becomes '10:30 AM', 'doctor smith' becomes 'Dr. Smith'.
Why does Inverse text normalization matter?
ITN is why usable transcripts do not spell out every number and address. It is locale-dependent (dates, currencies and phone formats differ by country) and one of the quiet quality gaps between transcription systems.
What terms are related to Inverse text normalization?
Closely related concepts: Punctuation restoration, Transcript editing — each has its own entry in this glossary.
Transcribe audio or video in 99+ languages — speaker labels, word timestamps, captions.
Transcribe a file →