#285 Real-Time Speech AI and Accent Translation with Sanas CEO Sharath Narayana

Sharath Narayana, CEO and Co-Founder of Sanas, joins SlatorPod to talk about the evolution of real-time speech AI, the rise of accent harmonization, and why voice will become the next major enterprise interface.Sharath traces his journey from engineer to entrepreneur, including the founding of Observe.AI before launching Sanas alongside Stanford researchers focused on solving low-latency speech processing.The CEO explains that Sanas initially operated as a speech lab focused on improving human understanding in conversations. The company developed multiple algorithms covering noise cancellation, speech enhancement, accent harmonization, and language translation before discovering that enterprises were most willing to pay for accent-related technology.Sharath emphasizes that Sanas’ accent technology is not designed to erase identity, but to improve clarity and reduce friction in customer interactions. He says enterprises, especially contact centers, adopted the technology to improve first-call experiences and reduce mistrust between agents and customers.He also discusses Sanas’ focus on on-device AI infrastructure rather than cloud-only deployments, where running speech AI locally improves latency, protects data sovereignty, and lowers compute costs.Looking ahead, Sharath says Sanas is preparing broader launches for real-time language translation, universal accent translation, and developer SDKs that will allow third parties to build voice applications on top of the Sanas platform.