Summary
- Protects calls against voice-based fraud on both sides
- Customer voice change detection
- Deepfake detection
- Agent voice authentication
- Managed under Team Settings >>> Products >>> Voice Security, team-wide, each toggled independently.
Voice Security is a set of Krisp Call Center AI features that help protect calls against voice-based fraud. It uses voice analysis to flag unexpected voice changes and AI-generated speech on a call, on both the customer side and the agent side, so your team can act on possible impersonation or agent substitution.
Voice Security settings are managed by an Admin and apply to the whole team. Each feature can be turned on or off independently.
Info
To get started with Krisp Voice Security:
- New customers: contact our Sales team at sales@krisp.ai.
- Existing customers: contact Customer Engineering at support@krisp.ai.
What Voice Security includes
Voice Security currently includes three features:
| Feature | What it does |
|---|---|
| Customer voice change detection | Detects and informs agents when an unexpected voice change is identified on the customer side of a call, helping agents spot possible impersonation. Works with Krisp AI Voice Translation. |
| Deepfake detection | Analyzes customer speech in real time to identify whether the agent is talking to a human or an AI-generated voice. Flags the call, collects analytics, and can alert the agent when synthesized speech is detected. |
| Agent voice authentication | Automatically detects and flags agents when an unexpected voice change is identified. Agent voice biometric data is used to monitor voice consistency across calls, helping you detect agent substitution. |
Where to manage Voice Security
All Voice Security settings live in one place in the Admin Portal. Go to Team Settings >>> Products >>> Voice Security. From there, you can turn each feature on or off.
Customer voice change detection
When this setting is on, Krisp monitors the customer side of the call and informs the agent when an unexpected voice change is identified. This gives agents an early signal of a possible voice change during a live call.
For a detailed breakdown of the feature, see the Customer voice change detection article.
Deepfake detection
When this setting is on, Krisp analyzes customer speech in real time and flags the call when it identifies an AI-generated voice. Admins can set a detection delay, enable real-time alerting, and choose whether a whisper is played when a synthesized voice is detected. Krisp does not decide how to handle the call: agents follow your organization's procedures.
For detection settings and what agents see during a call, see the Krisp Deepfake Detection article.
Agent voice authentication
When this setting is on, Krisp uses agent voice biometric data to monitor voice consistency across calls and flags agents when an unexpected voice change is identified. This helps you detect agent substitution.
For setup and details, see the Agent voice authentication article.