CNTXT AI Launches Munsit Emirati TTS to Perfect Localised Voice Tech

Voice technology is moving fast. We have seen global giants like Google, OpenAI and ElevenLabs roll out increasingly polished speech systems over the past year. Yet, for all that progress, one thing has often felt slightly off in our region. The tone might be clear, the pacing smooth, but the dialect? Not quite spot on.

This week in Abu Dhabi, CNTXT AI introduced what it says is the most accurate native Emirati text-to-speech model to date. Called Munsit Emirati TTS, the system is designed to generate real-time, human-like Emirati Arabic for both enterprise and consumer use. And yes, it’s aimed squarely at closing a gap that many in the UAE have quietly noticed for years.

In simple terms, text-to-speech (TTS) technology converts written words into spoken voice. But doing that in a way that truly reflects how people speak, with the right rhythm, cultural nuance and local expressions, is a different ball game. Arabic itself presents technical challenges for AI systems, and regional dialects add another layer of complexity. Emirati Arabic, in particular, has often been underserved.

Munsit Emirati TTS is built to respond instantly in spoken Emirati Arabic, whether that is through a bank’s call centre, a government service hotline or a digital assistant embedded in an app. The idea is to make these interactions feel less robotic and more natural. Not a scripted monologue, but something closer to a real conversation.

According to CNTXT AI, blind testing with Emirati and Arabic-speaking listeners showed that 93 per cent preferred Munsit Emirati over leading global models when it came to naturalness, emotional expression and dialect accuracy. That’s a strong claim in a market where international players dominate the headlines.

Mohammad Abu Sheikh, Founder and CEO of CNTXT AI, commented that voice is becoming more than just a tool for interaction. “Voice is no longer just an interface, it is becoming part of how services express identity,” he said. He noted that the region has long depended on imported systems that did not fully reflect local communication styles, adding that building technology aligned with the way people actually speak can influence trust and engagement.

Shameed Sait, AI Director at CNTXT AI, pointed out that most voice systems were not originally designed for Arabic, “and certainly not for Emirati”. He explained that the breakthrough was not only generating speech, but capturing rhythm, tone and cultural context, and making it work reliably at scale in real-world environments.

The enterprise angle here is key. The model is designed for sectors such as banking, telecoms, government services and digital platforms, where thousands, sometimes millions, of voice interactions take place every day. Organisations deploying AI-driven voice systems have reported cost reductions of between 20 and 40 per cent, along with faster response times, especially in high-volume contact centres. For companies watching their operational budgets, that is definately not a small incentive.

From my side, having spent years listening to founders across the MENA region talk about localisation, this move feels like it was only a matter of time. I remember speaking to a startup team in Dubai who told me their biggest headache was making their chatbot sound “less foreign”. It was, as they said, a bit of a faff trying to tweak global models to reflect Emirati personality. An off-the-shelf American-accented Arabic voice just doesn’t cut it when you are serving citizens in Sharjah or Al Ain.

That said, technology alone is never the full story. Adoption will depend on how easily organisations can integrate the system into existing infrastructure and how it performs under pressure. Contact centres are unforgiving environments, and even minor latency issues can frustrate customers.

Still, there is a broader shift happening. Across the UAE and the wider MENA region, companies are moving away from English-first or neutral voice systems towards tools that reflect local identity. And believe it or not, how a service sounds can shape how it is perceived, whether it feels distant or familiar, transactional or trustworthy.

I reckon this focus on dialect-specific AI is long overdue. If voice is becoming a core interface for digital services, then it needs to sound like the people it serves. With Munsit Emirati TTS, CNTXT AI is betting that authenticity in speech will be just as important as speed or efficiency. For a region investing heavily in AI, that feels like a logical next step… and perhaps a necessary one.

🚀 Got exciting news to share?

If you're a startup founder, VC, or PR agency with big updates—funding rounds, product launches 📢, or company milestones 🎉 — AraGeek English wants to hear from you!

✉️ Send Us Your Story 👇

https://en.arageek.com/send-us-your-story-or-a-news-tip