Real-time listener-side voice AI improves comprehension across meetings, contact centers, and voice agents

BERKELEY, Calif.: Krisp today introduced Listener-side Accent Conversion, a real-time, voice AI technology designed to improve how accented English is understood in live conversations. The system operates on-device in real time and is built to support human collaboration, customer experience operations, and voice AI agents.

For years, voice technology has focused on improving how conversations sound or how they are recorded. Noise cancellation reduces background interference. Transcription captures what was said.

Even with clear audio and accurate transcripts, understanding can still break down in real time, particularly across accents.

Listener-side Accent Conversion now addresses that gap directly.

Instead of modifying how a speaker talks, the system adapts incoming speech for the listener in real time, clarifying commonly misheard sounds while preserving the speaker’s natural voice and tone. Only the listener hears the adapted audio.

A Constraint Shared by Humans and AI

Accent variability affects communication across multiple environments:

• Global meetings: Participants repeat themselves, slow conversations, or miss context.
• Contact centers: Agents working across diverse accents experience increased repetition, longer handle times, and higher cognitive load.
• AI voice agents: Accent diversity reduces recognition accuracy and automation performance.

As voice becomes a primary interface for work and customer interaction, comprehension is emerging as a systems-level requirement, not just a personal challenge.

Founder Perspectives

“I’ve spent more than 20 years working in tech with an Armenian accent. I know what it feels like to repeat yourself on a call, or to see someone concentrating on your pronunciation instead of your idea. Over time, that changes how freely people speak. We built Accent Conversion because communication should be about ideas, not decoding speech. If technology can remove that barrier in real time, conversations become clearer and more equal for everyone involved.” - Arto Minasyan, Co-Founder and President, Krisp

“In contact centers and AI systems, the strain isn’t abstract. Agents process multiple accents all day, often in a second language. That adds friction, time, and cognitive load to every interaction. Listener-side Accent Conversion addresses the problem at the point where speech is received, helping both humans and AI systems operate more reliably without asking anyone to change how they speak.” - Davit Baghdasaryan, Co-Founder and CEO, Krisp

For Meetings, CX and Developers

Listener-side Accent Conversion is generally available today for human-to-human meetings through Krisp’s Voice AI for Meetings application.

The same listener-side modeling is being integrated into Krisp’s Call Center AI platform, where it enhances what agents hear during live customer calls. By reducing repetition and cognitive strain, the technology is designed to support faster resolution and improved customer experience without requiring customers to change how they speak.

The technology will also be available through the Krisp SDK, enabling developers to embed the capability directly into their applications and Voice AI Agents.

With bidirectional Accent Conversion, Krisp now supports accent clarity on both sides of live conversations.

How It Works

• Processes incoming audio at the phoneme level
• Clarifies sounds commonly misheard across accents
• Runs locally on-device with <200ms latency, imperceptible the human ear
• Requires no transcripts or post-processing
• Stores no raw audio

Models are trained across diverse English accents and designed to improve intelligibility in global meetings, delivering the strongest results across Indian, Filipino, Latin American, African, and Chinese-Mandarin accents, while improving comprehension across many others. Accent coverage continues to expand.

Listener-side Accent Conversion is available today within Krisp’s Voice AI for Meetings application for Mac and Windows. Integration into Krisp’s Call Center AI platform and SDK availability are underway.

Learn more at https://krisp.ai/ai-accent-conversion/listener/.

About Krisp

Founded in 2017, Krisp pioneered the world’s first real-time Voice AI models and software. Krisp’s Voice AI technology enhances digital voice communication through audio cleansing, noise cancellation, accent conversion, live speech-to-speech translation, and agent assist. Krisp applications are privacy-first, using industry-standard security practices across all audio hardware configurations and applications that support digital voice communication. Today, Krisp is deployed on over 200 million devices, has transcribed over 120 million calls, and processes over 80 billion minutes of voice conversations every month, helping businesses harness the power of voice to unlock higher productivity and deliver better business outcomes.

Fonte: Business Wire