Scribe V2 Feature Update

Scribe v2 has been significantly updated to give developers finer control over privacy protection, multilingual accuracy, transcription quality, and domain-specific precision.

The key updates are as follows:

• Entity Redaction: Sensitive personal information such as names, credit card numbers, and social security numbers is automatically detected during transcription and masked in a form such as [REDACTED] or [CREDIT_CARD], strengthening data security.
• Improved Indic-English code-switching: When mixing Indian languages such as Hindi and Telugu with English, the English portions are accurately transcribed in Latin script rather than the local script, reflecting how people actually communicate.
• No Verbatim mode: Filler words like "um" and "uh," repeated phrases, and stammering are automatically removed, providing a clean transcription output without additional editing.
• Expanded keyterm prompting: The number of supported keywords has increased tenfold, from 100 to 1,000, maintaining high accuracy even in technical fields with a lot of specialized terminology.

These features are now immediately available through the Scribe v2 API and UI.