Sanas builds a patented Real-Time Speech Understanding Platform that applies generative AI to transform speech in real time and remove background noise. The company’s technology is used for accent harmonization and to enhance clarity and inclusion in multilingual interactions. Sanas was founded in 2020 and is led by co-founders CEO Sharath Keshava Narayana and CTO Shawn Zhang. The platform has been deployed by early partners such as Alorica, which reports measurable operational improvements tied to the technology (for example reduced Average Handle Time, fewer call transfers, higher conversion rates and strong agent adoption). Sanas holds a patent on its real-time AI speech understanding technology and continues to iterate with strategic customers to refine applications across customer experience scenarios. Financially, Sanas recently closed a new funding round (see deal details). Sanas develops an AI "accent translation" engine that filters and converts accented speech in real time while preserving a speaker’s natural voice. The technology is deployed primarily in call centers and has seen traction with large customers including Assurant, Alorica, ERC and IGT. Sanas runs its system on premises (not in the cloud) and says the method is in the patent process; the product is intentionally not available to consumers to limit potential misuse. The company says the engine works across languages and is already being used to smooth accents in regions such as Japan, China and South Korea. Sanas announced a strategic partnership with Alorica, which will roll the tech out to 100,000 employees and 250 enterprise customers globally. The company is coming out of stealth, plans to expand within the call-center vertical and to develop other enterprise use cases such as video-call plug-ins and voice-based interactive services. Sanas develops machine-learning software that recognizes a speaker’s speech and outputs the same words with a chosen accent in near real time, running locally without requiring the cloud. The system slots into the operating system’s sound stack so it works out of the box with most audio and video calling tools. Sanas is running pilot programs with thousands of users across the U.S., U.K., the Philippines, India, Latin America and other locations; testing has suggested reductions in error rates and increases in call efficiency, though no deployment metrics are yet available. By the end of the year the company plans to support American, Spanish, British, Indian, Filipino and Australian accents. The current output can sound noticeably more artificial and retains the speaker’s character and cadence only partly, but the company says the engine improves with usage. CEO Maxim Serebryakov frames the mission as making communication easy and free from friction so people can speak confidently and understand each other.

Total raised

$103M

Funding rounds

3

Latest round

Series B

Latest activity

Apr 2025