Krisp today rolled out customer-side Accent Conversion, a real-time, inbound Accent AI capability to improve agents' comprehension during live customer calls.
The technology operates in real time, adapting what the agent hears to enhance intelligibility while preserving the customer's natural voice, emotion, and context of the conversation. Only the agent hears the adapted audio.
Built for global contact centers, the system can handle diverse accent patterns within a single, dynamically adapting model, without requiring accent selection, configuration, or workflow changes. It operates seamlessly within the agent's Krisp app.
"In contact centers, the strain isn't just on customers; it's on agents who are often working in a second language and processing multiple accents all day," said Davit Baghdasaryan, CEO and co-founder of Krisp, in a statement "When agents have to work harder just to understand what's being said, cognitive load increases and performance suffers. Customer Accent Conversion reduces that effort in real time, helping agents stay focused, resolve issues faster, and deliver better customer experiences without asking customers to repeat themselves or change how they speak. The downstream impact is tangible: reduced handle time, faster resolutions, and improved satisfaction on both sides of the call."
The technology is built for enterprise security and scale. Processing happens locally on the agent's device, and raw audio is never stored or routed through external servers. Krisp operates as a virtual microphone and speaker, enabling deployment across major contact center-as-a-service platforms and softphone applications without workflow changes.