Microsoft Teams will soon clone your voice in 9 languages

Language barriers in corporate video calls usually boil down to an awkward choice: struggle in broken English, or listen to a robotic, monotone translation voice that drains all personality from the room.

Microsoft wants to fix that. At its Ignite 2024 conference, the company unveiled Interpreter in Teams, a new real-time speech-to-speech translation tool that can simulate the speaker's own voice in another language.

Set to roll out in early 2025 exclusively for Microsoft 365 subscribers, the feature aims to make cross-border meetings feel far more natural.

Sounding like yourself in 9 languages

According to Microsoft CMO Jared Spataro, Interpreter allows attendees to sound like themselves even when speaking a language they don't know a word of.

At launch, the tool will support nine languages: English, French, German, Italian, Japanese, Korean, Portuguese, Mandarin Chinese, and Spanish.

Instead of hearing a standard corporate synthesized voice, your foreign colleagues will hear your tone, inflection, and cadence delivered in their native tongue in real time.

Sure, machine translation still fumbles colloquialisms and cultural nuances compared to top-tier human interpreters. But the enterprise appeal is massive. According to Markets and Markets, the natural language processing sector, including translation, is projected to hit $35.1 billion by 2026.

Opting in to your own clone

Putting voice cloning into the world's most ubiquitous workplace software is an obvious security headache.

Deepfakes have caused chaos this year, with impersonation scams costing victims over $1 billion last year, according to FTC figures. Just earlier this year, criminals staged a convincing fake Teams meeting with C-level executives, tricking a firm into wiring $25 million. OpenAI previously withheld its Voice Engine tool from the public precisely because of these safety concerns.

Microsoft is treading carefully with guardrails. The company stated that voice simulation will never run by default; users must explicitly give consent via an in-meeting prompt or enable it manually in settings.

Microsoft also emphasized that Interpreter does not store biometric data, will not inject artificial emotions beyond what is naturally in the speaker's voice, and can be switched off at any time.

Why it matters

For global workers, this could remove one of the most frustrating barriers in multinational collaboration. Talented employees who struggle with conversational foreign languages can finally speak freely in their native tongue without losing personal presence or sounding like automated IVR systems.

For the enterprise software race, Microsoft is turning speech cloning from a consumer novelty into a baseline corporate productivity feature. While Meta tests automated dubbing on Instagram Reels and ElevenLabs sells multilingual speech generation to developers, Microsoft is deploying this straight into the daily workflows of millions of companies. That puts serious pressure on Zoom and Google Meet to catch up.

We think Microsoft's implementation looks sensible on paper. Constraining the clone to live meeting participants with explicit consent is far safer than offering an open-ended audio generator.

Still, the risk isn't zero. If an account is compromised, a bad actor feeding prerecorded malicious audio through an authorized profile could easily mislead unsuspecting teammates. IT administrators will need to update internal verification protocols before letting staff rely on synthetic voices for sensitive decisions.