Names Entities Recognition

Our platform is able to recognize a fair number of named entities. Once spotted, it is possible to have this entity be redacted, depending on the transcription settings you chose. Note that by default, sensitive entities will be redacted.

In the transcript, you will find the Entity code in place of what was said. The audio file is also redacted with a jazzy music.

If you're looking for the named entities recognition support for our Stream API, please check the dedicated documentation.

We strongly recommend to at least configure redaction for Credit Card and IBAN numbers, whatever your use case is. Other entities should be redacted depending on your use case in order to be compliant with GDPR.

For a custom configuration of which entities have to be redacted, you can choose among the following ones:

DescriptionLanguage availability
Misc series of numbers,
Please note that a single number won't be redacted, for example: "I need 213 units", but 2 or more consecutive numbers will be, for example "The value is 752 634".
Credit card number,
if close enough to the card number, date of expiration and CVV will also be spotted.
IBAN bank account number,
for example as FR76 xxxx xxxx xxxx xxxx xxxx xxx.
Phone number.
ENEDIS delivery point,
specific to the French energy market.
Date entities,
for example tomorrow, monday 12th...
Time of a day,
for example noon, 07:00...
A period of time,
of less than 24 hours, for example 2 seconds, 4 hours...
First name.
Last name.
Spelled first name.
Spelled last name.
E-mail address.
Date of birth,
for example November 5, 1955.
Street path,
for example 22 rue Ferrandière, 3 Abbey Road...
Other location element,
for example floor number.
U.S. states.
Body organ.
such as a company name.
A product name.
An amount of money,
for example 15 euro, 16 dollars, 17 rubles...
A quantity,
for example 12 oysters, 2 kilometers, 85 kg...