Krisp has announced the launch of Listener-Side Accent Conversion, a voice AI technology for improving the intelligibility of accented English in digital conversations, such as those in meetings, customer experience (CX) settings, and voice AI tools. Unlike other accent conversion tools, Krisp’s new tool does not alter the voice of the person talking, as it is implemented on the listener’s end, converting the speech in real time for easier understanding. It works directly on the device with low latency, which means that the latency is less than 200 milliseconds, in order for the conversation to be had smoothly without the need for transcription or storage.
Also Read: LivePerson Introduces Syntrix Platform to Evaluate and Train Enterprise AI Agents
It works at the phoneme level, which means that it has a deeper understanding of the types of sounds that are commonly misheard, and this could be useful in reducing friction in the workplace, especially in the global world in which one might have to deal with people of different accents and languages. According to Krisp, the variability in accents often leads to unnecessary explanations, longer handling time in contact centers, and less accurate voice AI technology. This technology promises to improve productivity and cognitive ability of the users, as well as the overall accuracy of voice-based AI technology. This feature is already available in the company’s Voice AI for Meetings app and will soon be made available in the company’s Call Center AI platform and SDK, which would allow the company and other developers to incorporate the feature into their technology.