AI accent translators remove laughs, whispers, and inflections. For Wipro, UnitedHealth, and Wells Fargo, the flat voice risks alienating customers. Pew: 74% prefer human agents.
The customer joked about Friday the 13th while filing an auto insurance claim. The Wipro call-centre representative laughed. The AI accent translator from Sanas stripped the sound before it reached the customer. Five seconds of silence followed, then a flat “Yes I am here.”
The tool that converts Indian and Filipino accents into American English cannot relay a laugh, a whisper, or a change in tone. Silicon Valley startups like Sanas sell the software to bridge the gap between regional speech patterns and the polished English expected at call centres. Companies across banking, insurance, and outsourcing have adopted it. Wipro uses the tool when processing claims for insurance clients. UnitedHealth Group deploys it in its customer service arm, which handles millions of medical and billing calls annually. Wells Fargo runs in-house centres alongside outsourced ones that use similar translators. American Express, Teleperformance, and Alorica also employ the technology.
Teleperformance, the world's largest call-centre operator, said Sanas raised its Net Promoter Score by 26% across 42,000 Indian employees. The promise is straightforward: hire cheaper talent from smaller towns and deliver a voice that sounds local. Accent complaints drop. Calls shorten. NPS improves.
The limitation is harder to measure. The translator flattens every vocal signal into the same standard cadence. A hesitation, a giggle, an angry downshift – all get recoded into a mid-range monotone. Customers report the voice sounds human but flat. Some suspect they are speaking to a bot. A 2024 Pew Research survey found 74% of Americans prefer speaking with a human agent for complex issues. When the voice sounds artificially uniform, trust erodes faster than if the accent were unfamiliar.
For Wipro, the exposure cuts two ways. Its call-centre staff use the translator when handling insurance and banking clients. The company also licenses similar AI tools to its BPO clients. A misfire that alienates customers could push clients to demand offline handling or renegotiate penalties. Margins in the IT-services business are already thin. Wipro's Alpha Score is 46 out of 100, reflecting mixed market sentiment on its ability to manage technology transitions.
UnitedHealth Group fields calls about medical claims, billing disputes, and coverage questions. A misread inflection on a claim denial call – where tone matters more than the words – could escalate a complaint into a regulatory issue. The company faces scrutiny on service quality from state insurance commissioners. Its Alpha Score sits at 47.
Wells Fargo uses accent translation in mortgage modification and banking support calls. Losing trust on a call about a loan restructuring is an expensive problem. The bank has spent years rebuilding its reputation after the fake-accounts scandal. A voice that sounds like a bot undercuts that effort. Its Alpha Score of 63 is the highest among the three firms.
More companies will adopt accent translation anyway. Sanas raised $80 million in 2024 and signed multi-year contracts with three of the top five global BPOs. The challenge is adding the emotional layer back into the pipeline. A few startups train models to detect sentiment from pitch and speed, then re-inject that information as a label or a tone shift. None has proved the system works at scale.
Teleperformance said the translator improved its NPS by 26% across 42,000 employees. Customers are not laughing.
Drafted by a large language model from the source reporting linked above, then screened by automated publishing checks. It is not read by a journalist before publication. Some articles cite our Alpha Score. Verify prices and figures against the original source. Educational coverage, not personalized advice.