V-Blaze and V-Cloud Online Help (May 2020)

Redaction

By default, V‑Blaze produces transcripts containing all information present in the available speech. In some cases, it may be desirable for certain spoken phrases to be removed from the final transcript. For example, it is often necessary to remove credit card numbers, social security numbers, and profanity from transcripts. Redaction allows targeted masking of particular phrases and number patterns, and it can be applied to both the text transcript and the original call audio.

The speech transcription server does not redact text or audio by default. Furthermore, requesting redacted text does not automatically generate redacted audio.To access either feature, you must request redacted audio and text when you submit the audio data.The method used to request audio redaction depends on the interface used for submitting the audio data.