Lost in Translation: What Accent Tech Reveals About Bias, Clarity and Connection

In a Slack conversation among CX professionals, Luke Jamieson raised a simple question as part of a CX question of the week on the CX Accelerator: What are the real implications of accent reduction technology in contact centers?

The discussion quickly moved beyond the tech itself. It touched on ethics, identity and the subtle forces shaping how we interpret and trust the voices on the other end of the line. Natalie Perez, who joined the conversation, found herself reflecting on her own experience as a non-native English speaker. This wasn’t just a customer experience topic. It was deeply personal.

What Is Accent Reduction Technology?

Accent reduction, sometimes also (horribly) called accent neutralization, as in “neutralize the threat,” uses AI to alter a speaker’s voice in real time. The result is a more "neutral" or "standard" pronunciation that aligns with the accent associated with the customer base geographical location. This is often used in offshore contact centers to reduce perceived friction between agents and customers.

The reasoning seems straightforward: clarity improves experience. But clarity for whom? And at what cost?

Why It’s Gaining Ground

For decades, contact centers in countries like the Philippines, India and Colombia have fielded bias from customers who associate foreign accents with poor service. This isn’t about capability. It’s about perception. Even when agents deliver excellent outcomes, the sound of their voice can influence how they are judged.

In response, companies have begun implementing accent-modifying tools. One high-profile example is Teleperformance’s partnership with Sanas, a startup offering real-time voice modulation. The stated aim is to reduce miscommunication and improve satisfaction.

However, this raises a bigger question: Are companies solving a problem of communication, or responding to customer bias by masking their workforce?

The Trust Factor

Luke has been referencing research lately that shows poor audio quality impacts how believable a speaker sounds. One study from the University of Southern California and another from Australian National University both found that listeners trust messages less when audio quality is low — even when the content is identical.

It’s not a stretch to consider that accents might also influence trust in a similar way. Not because of what’s being said, but because of how it sounds. This opens the door to using tech for the wrong reason: to align voices with an arbitrary ideal of credibility.

Where Do We Draw the Line?

Some studies show that female voices are rated as more trustworthy or compassionate in certain contexts. Others link perceived generosity or authority to specific vocal tones or genders. If companies start tailoring voice presentation based on these traits, it’s easy to imagine a future where voices are modified not just for clarity, but to hit performance targets.

That is a slippery slope. If an agent's voice can be changed to sound more “neutral,”could it also be adjusted to sound more male, more senior or more empathetic? Could it be used to pretend a call has been escalated to a manager when it hasn’t?

Is This a Two-Way Street?

One intriguing thought: If customers can request clearer communication, could agents do the same? What happens if an agent can’t understand the customer and wants to apply a voice modifier to them?

This would likely cause outrage, but it highlights the one-sided nature of these tools. They cater to customers' preferences without addressing the agent experience or giving them similar control.

The Impact on Agents

Accent is not just a technical trait. It is part of identity. For many agents, their voice is tied to their personal story and sense of professionalism. When companies modulate that voice without care, they risk sending the message that the agent’s natural way of speaking isn’t good enough.

That can undermine confidence, motivation and even lead to higher attrition. In a space where engagement is already fragile, it’s a high cost to pay for perceived polish.

However, it could also go the other way because it might give agents confidence that they will not experience certain biases based on their accent.

What Should Companies Do?

Natalie and Luke agree that this isn’t about being anti-tech. It’s about being intentional with how technology is applied, as ICMI’s Erica Marois raised in the discussion. Accent tools may help with accessibility or severe communication barriers, but they shouldn’t be a default.

Instead, they suggest a more balanced approach:

• Start with listening. Pinpoint whether the issue is accent, audio quality or something else before reaching for AI.
• Invest in training. Teach agents and customers to communicate across differences. Empathy is a skill that improves both sides of the experience.
• Be transparent. Let agents know if their voice is being altered and give them a choice.
• Look beyond the numbers. Measure the impact on agent dignity and retention, not just CSAT or average handle time.
• Design for inclusion. The best CX is human-centric and tech-enabled, not tech-first.

Contact centers are where brand, empathy and humanity intersect. Accent reduction tools may help in specific cases, but they must be used with care. There is power in voice diversity, and progress should not come at the cost of authenticity.

In a world increasingly shaped by artificial intelligence, real connection still matters, and sometimes, that starts with letting people sound like themselves.