Krisp, renowned for its innovative noise cancellation application, has introduced a new artificial intelligence tool designed to convert various accents to American English in real time. This feature is integrated into Krisp’s existing desktop application, functioning as a virtual microphone that alters a user’s voice during video conferencing on platforms such as Zoom, Microsoft Teams, Google Meet, and Webex.
The AI Accent Conversion tool offers a conversion latency of 200 milliseconds, characterized by Krisp as an “imperceptible delay” in conversations, while retaining the user’s natural vocal tone. Demonstrations by the company suggest the technology effectively modifies speech to make it sound more American, although some users noted that the output had a slightly robotic quality.
Currently, the tool accommodates over 17 Indian dialects, with plans to expand its capabilities to include additional English accents, such as Filipino.
The initial rollout of this tool occurred within call center agencies before a wider public launch. Arto Minasyan, co-founder and president of Krisp, expressed in a press release, “As someone with an accent, I’ve often noticed that people struggle to understand me, even when my English is fluent. This isn’t about bias — it’s simply a reality of communication. In fast-paced meetings, even small misunderstandings can slow down decisions, cause repetition, and reduce overall efficiency.”
Currently, the AI Accent Conversion is being offered in beta form, with free users granted access for up to 60 minutes daily. A subscription service priced at $15 per month enables unlimited use for business clients.
Krisp is not alone in leveraging AI to modify speech accents. Teleperformance SE, another key player in the call center industry, has begun implementing an AI-driven tool aimed at “softening” the accents of Indian employees, as reported by Bloomberg. Nevertheless, the use of AI in this context poses complex challenges related to race and ethnicity, and AI-generated voice technology carries its own set of risks.
