UAE-Based CNTXT AI Launches On-Device Arabic Speech-to-Text System

CNTXT AI, an Abu Dhabi-based company focused on AI infrastructure, language models, and enterprise data systems, has launched Munsit Edge, an Arabic speech-to-text platform designed to operate entirely on-device without requiring a cloud connection.

Built on the company’s Munsit Arabic automatic speech recognition model, the system supports real-time transcription across major Arabic dialects using consumer hardware including smartphones, computers, in-car systems, and Internet of Things devices.

According to the company, Munsit Edge processes audio locally on devices rather than transmitting data to external servers, a setup intended to reduce dependency on cloud infrastructure while improving privacy and response times.

CNTXT AI reported that the platform achieves an approximate 24% word error rate across major Arabic dialects while supporting real-time transcription with latency of around 150 milliseconds on standard mobile devices.

The system is being released through software development kits compatible with iOS, Android, macOS, Windows, and Linux, alongside on-premise deployment options and embedded Internet of Things integrations.