Published On: Fri, Feb 28th, 2020

Amazon Transcribe can now automatically redact privately identifiable information

Amazon Transcribe, a AWS-based speech-to-text service, launched a tiny yet critical new underline this morning that, if implemented correctly, can automatically censor your privately identifiable information from call transcripts.

One of a many renouned use cases for Transcribe is to emanate a record of patron calls. Almost by default, that involves exchanging information like your name, residence or a credit label number. In my experience, some call centers stop a recording when you’re about to sell credit label numbers, for example, yet that’s not always a case.

With this new feature, Transcribe can automatically brand information like a amicable confidence number, credit label number, bank comment number, name, email address, phone series and mailing residence and redact that. The apparatus automatically replaces this information with ‘[PII]’ in a transcript.

There are, of course, other collection that can mislay PII from existent documents. Often, though, these are focused on information detriment impediment collection and aim to keep information from leaking out of a association when we share papers with outsiders. With this new Transcribe tool, during slightest some of this information will never be accessible for pity (unless, of course, we keep a duplicate of a audio).

In total, Transcribe now supports 31 languages. Of those, it can register 6 in real-time for captioning and other use cases.

About the Author