Don't worry if your English has too much of an Indian accent: startup Krisp launches AI conversion to authentic American in one click

March 27, 2012 - According to foreign media outlet TechCurnch, audioStartups Krisp On Wednesday, the launch of a AI Accent Conversion Function, users can change their accents during calls. The first version supportslocalizeEnglishAccent Conversion to American English.

Krisp stated that the featureDoesn't change the speaker's voiceThe Krisp feature is now available in Beta for corporate environments, adjusting only the phonemes to match American pronunciation. The feature is currently being tested in corporate environments and the Beta version is live on the Krisp desktop app, allowing users to enable it at any time before or during a call.

Company co-founder Aalto Minassian says the feature grew out of his personal experience with communication. "Even though my EnglishfluentBut many people stillcan't understand what's going onMy accent. We thought that changing the accent might help people communicate more smoothly. We started working on this two years ago and now we're finally launching Beta."

The outlet's tests, however, found that the conversion didn't sound natural enough, and even occasionally missed words.Krisp explains that there are stillIn Beta, the model will be optimized in subsequent updates.

Krisp says that in enterprise testing, the feature allowed theSales Conversion Rate Improvement 26.1%The revenue per book increased by 14.81 TP3T.

The company chose to support Indian accents first due to IndianIn the global STEM field(1AI note: Science, Technology, Engineering and Mathematics)A high percentage of workers in the. There are plans to add more accent options in the future, includingPhilippine accent.

Krisp stated that the model is based onThousands of speech samples trainingthat covers different accents and dialects and uses data from its meeting assistant with the user's consent. Minasyan adds that another big advantage of the feature is thatNo need to collect user voice data in advance, but rather the ability to create personal voice models in real time.

statement:The content of the source of public various media platforms, if the inclusion of the content violates your rights and interests, please contact the mailbox, this site will be the first time to deal with.
Information

ChatGPT's New Image Generator Goes Live, Ghibli-Style Images Brush Off Copyright Controversy

2025-3-27 19:05:57

Information

Huawei ModelEngine Full Process AI Development Toolchain Open Source

2025-3-27 19:14:24

Search