Inverse Text Normalization

caution

This feature is only available for Sommers model when language is "ko".

Converts expressions like letters, numbers, abbreviations, emails, phone numbers, dates, and units into cleaner forms.

Examples:

• Korean text with numbers and abbreviations can be normalized to numerals and Latin letters.

Enable by setting use_itn: true (default true).

Usage

In the POST Request body, set use_itn: true to enable normalization.

POST example: Example

Unified example (presets): Example