Sanas.AI Inc., a real-time accent translation startup that helps non-native speakers speak more clearly, said today it has raised $65 million in a Series B round of funding.

The round brings Sanas’s total amount raised to north of $100 million. It was led by Quadrille Capital and saw participation from investors such as Insight Partners, Quiet Capital and DN Capital, as well as the global digital services provider Teleperformance, which says it has also become a strategic partner.

The startup, which is headquartered in Palo Alto, knows a thing or two about the challenges faced by nonnative speakers, for its four founders are all first-generation immigrants to the U.S.

Maxim Serebryakov, Shawn Zhang, Sharath Keshava Narayana and Andrés Pérez Soderi told TechCrunch in a 2021 interview they were inspired by one of their friends, who worked at a call center in Nicaragua serving U.S. customers. Despite being fluent in English, their friend endured terrible abuse from U.S. callers who could not understand his accent.

Sanas sets out to overcome this problem. Its artificial intelligence-powered software is primarily aimed at call centers and others who need to spend lots of time on the phone, and it works by letting users set their preferred accent, such as U.S. English, British English and the like. Compatible with more than 800 communications-based applications, it works by listening to the user’s speech, recognizing the words they’re saying, then spitting out the same words in a different accent, while retaining their original voice.

According to the startup, its technology can support multiple languages, including French, Spanish, Italian, German, Chinese and Japanese. It’s worth pointing out that Sanas doesn’t translate one language into another, but rather modifies the user’s pronunciation in real-time, ensuring they can communicate in an accent that’s understood by the other speaker, without any latency.

As such, it’s completely different from speech translation, and much more similar to something like voice cloning. That puts it up against  startups such as Tomato.ai Inc., which positions its software as an AI-powered “accent softening” tool, and the synthetic voice startup Eleven Labs Inc.

The software performs its accent translations locally rather than in the cloud to reassure customers that it’s not storing their voice data. It says customers retain full control over their data at all times.

Serebryakov, the company’s chief executive officer, says accent-based bias has become a global challenge. He wants to make it so that people can easily understand each other, regardless of their origins. “Our patented technology allows people to maintain their voice and accent, while also helping businesses reduce costly, time-intensive hiring cycles and minimize the likelihood of attrition,” he said.

Constellation Research Inc. analyst Holger Mueller agreed, saying he has personal experience of the problems caused by speaking accented English.

“While the world speaks English, not everyone speaks ‘English’ English or even American English. In fact there are more accents and flavors of English than there are countries that speak it,” he said. “This all ends up making communication harder, and it’s something that arguably matters in call centers more than anywhere else, where customers are trying to communicate problems and agents aim to communicate the solutions. Sanas offers a promising AI-powered solution to eliminate these issues.”

Sanas reckons that its software has become a big hit. According to Narayana, it’s being used by “hundreds of thousands” of customer service agents across 39 countries, spanning industries such as financial services, logistics, healthcare, hardware manufacturing and business process outsourcing.

Looking forward, Sanas wants to expand its platform by adding new speech-to-speech algorithms to cover more accents, so it can enter new geographic markets and expand to use cases beyond call centers. The funds will help to build out its infrastructure and teams to support this expansion, and also improve its technology capabilities, which extend to noise cancellation.

Image: SiliconANGLE/Meta AI