Cobalt
Privacy Screen TM

Share your data while safeguarding customer trust.

Using the latest advancements in machine learning, Cobalt Privacy Screen engine detects, redacts & replaces personally identifiable information based entirely on context. This allows you to share your data while also safeguarding customer trust

Cobalt Privacy Screen

  • Identify over 50 types of personally identifiable information
  • Process 70,000 words/second in 8 languages, and is optimized for the idiosyncrasies of day-to-day conversations.
  • Hit a 99.5% accuracy rate on data like transcripts, and health records. 
  • Identify and redact text in 8 languages (with more on the way!) 
  • Allow you to retain full ownership of your data, it never leaves your infrastructure.

Cobalt Privacy Screen is available as a command-line tool that can operate on locally available files. Alternatively, it can run as an integrated part of Cobalt’s streaming speech-to-text engine to produce redacted transcripts of your live audio in real-time.

We know that privacy, and what is or isn’t considered personal information, can get complicated. Our team ensures that our system works in compliance with major legislation like GDPR, CPRA, and HIPAA, to help provide you with peace of mind. 

Cobalt VoiceBio TM

Highly Secured with Voice Biometrics and Authentication

Cobalt VoiceBio is a biometrics analyzer that can help you identify who is speaking, by enrolling voiceprints of known customers and employees. It can be combined with a passcode or other credentials for a powerful multi-factor authentication tool.

Authenticate your customers, employees and partners faster with Cobalt Voice Biometrics.

Improve customer experience by reducing the burden on customers to re-authenticate with cumbersome questions, especially again after a transfer. Can be deployed embedded on your device, on-premise or in your private cloud.

Security and Privacy

Cobalt VoiceBio can be configured to run on a local server, so users don’t have to worry about their voiceprints ending up in the hands of a third-party speech company.

Use Cases / Examples

Cobalt VoiceBio has two main uses:

Speaker Identification

In this case, we have a population of known customers or callers.  We create voiceprints for everyone. When a new audio stream is received, Cobalt VoiceBio compares it against the voiceprints on file and reports on the best match.

Speaker Authentication

With speaker authentication, a user creates a voiceprint through a short enrollment. Later, Cobalt VoiceBio can verify whether a new audio stream is from the same speaker, and show a red or green alert to a contact center agent to indicate how many verification questions they need to ask. This greatly reduces the time spent authenticating, particularly after transfers or calls back.

Challenge

A home automation company wanted to integrate spoken language controls & voice biometrics in their devices. However, most off-the-shelf solutions are either too expensive or not customizable for their specific needs.

Solution

They came to Cobalt to develop a roadmap for a solution that they could develop themselves or in partnership with Cobalt based on their requirements.

Result

Engaging With Cobalt has allowed them to have an option to own a solution tailor-made for their industry & needs

USE CASE

Cobalt Voice Biometric

Cobalt Speech Intelligence TM

Speaker Characterization by Voice

Cobalt Speech Intelligence is a speaker classification tool that can support modeling of various characteristics of the speaker, from demographic information (such as age, gender, regional accent) to current state of mind (such as emotion, fatigue, or intoxication).

This technology can provide you with valuable real-time information about your users to help ensure successful experiences. For example, Sentiment and Emotional/Agitation Analysis can provide customer support agents with real-time feedback mechanisms to trigger next possible actions based on customer sentiment to ensure a successful interaction. Or stress/fatigue detection can help supervisors identify potential employee burnout.