Teleperformance uses AI to wipe Indian accents in real time

If you've ever called tech support, you know the routine: you're frustrated, and the agent on the other end has a thick accent you're straining to decode. Now, the world's largest call center wants to make that dynamic vanish using real-time voice filters.

Teleperformance, the Paris-based outsourcing powerhouse behind customer support for Apple, Samsung, and TikTok, is deploying AI to "soften" and neutralize the accents of its India-based staff.

Earlier this year, the company quietly invested $13 million in Sanas AI, an accent-translation platform, snagging exclusive rights to the tech. The software uses machine learning to scrub non-native speech patterns on the fly.

A seamless filter for human agents

According to Teleperformance deputy CEO Thomas Mackenbrock, the system works with virtually zero latency. He told Bloomberg that Indian representatives can sometimes be difficult for native English speakers to understand, leading to longer calls and frustrated callers.

Mackenbrock argues that smoothing out the accent creates instant intimacy, drives up customer satisfaction scores, and slashes average handling time. He framed it as a win-win that keeps real humans in the loop during an era of rampant automation.

It is similar to techniques recently seen in Hollywood, like the AI accent adjustments deployed in the 2024 Oscar-nominated film The Brutalist. But applying it to real-time customer service comes with far trickier optics.

Investors give it the cold shoulder

While management pitched the feature as a breakthrough, Wall Street was deeply unimpressed.

After Teleperformance revealed on an investor call that it plans to pump up to $104 million into various AI technologies, its stock plunged 11 percent. That came despite the firm reporting a massive $11 billion in consolidated revenue for 2024.

JPMorgan analyst Sylvia Barker noted that the company missed expectations despite claiming higher-than-anticipated synergies. Investors appear uneasy about the ballooning tech bill, and likely nervous about the ethical backlash of digital whitewashing.

Why it matters

Teleperformance is trying to pull off a delicate high-wire act: protecting its core human-outsourcing business by patching human flaws with artificial intelligence.

For callers, clearing up linguistic barriers might shave a minute off a billing dispute. But running human voices through a synthetic filter feels inherently dystopian. If an agent's audio glitched or sounded robotic, customers would feel misled rather than helped.

Strategically, this looks like an expensive half-measure. As frontier voice models get cheaper and more conversational, enterprise clients like Apple and TikTok will soon face a simpler choice: keep offshore human teams or replace the tier-one queue entirely with autonomous voice bots.

Spending $104 million to bolt an AI mask onto a human call center worker feels like a stopgap before full automation arrives. Teleperformance's 11 percent stock drop suggests investors already see through the disguise.