Telus Alters Call-Agent Accents With Real-Time AI
Telus is using AI to modify offshore call-centre agents' accents in real time, according to reporting by iPhone in Canada and The Globe and Mail. The technology, supplied by Tomato.ai and deployed through Telus Digital, converts speech on live calls to reduce what the company describes as "accent-related friction."
Labour groups have called the practice deceptive and urged regulators to require disclosure to customers. Rogers and Bell told The Globe and Mail they have no plans to adopt similar technology, signalling divergence across Canada's telecom sector.
How the technology works
The system combines automatic speech recognition, accent conversion models, and neural vocoders to process audio with minimal delay. The software smooths phonetic patterns and prosody-the rhythm and intonation of speech-to reduce perceived accent differences.
Real-time voice conversion in call centres creates operational tradeoffs. Latency, naturalness, and robustness to background noise all compete for engineering resources. Noisy call-centre audio and multiple languages make the task harder than controlled settings.
What contact-centre teams should know
This deployment raises three immediate concerns for support operations: consent, transparency, and worker rights.
- Consent: Customers may not know their agent's voice has been altered, raising questions about disclosure obligations.
- Transparency: No public details yet on how Tomato.ai's model handles edge cases or what happens when the system fails.
- Worker rights: Voice-altering technology affects how agents are perceived and evaluated, with implications for performance metrics and job security.
Regulatory guidance on mandatory disclosure remains absent in Canada. That gap will likely close quickly given the public backlash.
What to watch
Canadian federal authorities may issue guidance on disclosure requirements for voice-altering AI in customer-facing services. Tomato.ai has not yet published technical details on latency, model architecture, or failure modes.
Other large contact-centre operators will face pressure to disclose whether they use similar tools and what safeguards they've implemented. Competitor silence may itself become a competitive signal.
For teams managing AI for Customer Support operations, this case demonstrates how Voice Modulation technology can move faster than policy. Building transparency into systems now-before regulators mandate it-will matter.
Your membership also unlocks:
