Abu Dhabi’s TII launches three AI models built for Emirati language

Falcon-Emirati, Falcon-ASR and Falcon-OCR-Arabic bring dialect recognition, speech transcription and document understanding to Arabic-language AI

Staff Writer
TII launches Falcon-Emirati alongside new Arabic AI models
Image: Emirates News Agency (WAM)

Article summary

AI Generated

Abu Dhabi's Technology Innovation Institute has launched three AI models built around Emirati Arabic: a dialect language model, a multilingual speech recognition system, and an Arabic document understanding tool. The releases expand TII's Falcon family into language, speech and visual content processing, with each model claiming benchmark-leading performance against larger open-source systems.

Key points

  • Falcon-Emirati scored 84.83% on an Emirati Arabic benchmark, beating all open-source rivals
  • Falcon-ASR transcribes six languages including Emirati Arabic, despite a 1.6bn-parameter footprint
  • Falcon-OCR-Arabic extracts Arabic text and structured content from images and documents

Subscribe to our free newsletter to continue reading.

Newsletters

Abu Dhabi’s Technology Innovation Institute (TII) has released three artificial intelligence models targeting Emirati language, speech and document understanding, expanding its Falcon family into territory where Arabic AI has historically lagged.

The first, Falcon-Emirati, is a 7-billion-parameter language model trained specifically on the Emirati dialect, capturing its vocabulary, idioms, expressions and cultural context. Built on Falcon-H1-Arabic and developed using native Emirati content, Modern Standard Arabic material covering Emirati culture and heritage, and synthetic data, it scored 84.83 percent on Alyah, an Emirati Arabic benchmark covering everyday language, figurative expressions, heritage knowledge and poetry. TII said the model outperformed every Arabic and multilingual open-source model evaluated on that benchmark. It will be accessible through TII’s Falcon Chat platform.

The second model, Falcon-ASR, handles spoken language. The 1.6-billion-parameter system transcribes Emirati Arabic, Modern Standard Arabic, English, French, Spanish and Portuguese into written text, and timestamps each word within an audio file.

Despite its relatively compact size, TII said it outperformed larger models on Arabic and Emirati speech benchmarks, including a 30-billion-parameter multimodal system on Emirati speech specifically. Target applications include subtitling, meeting transcription, accessibility tools and voice-enabled services.

The third, Falcon-OCR-Arabic, extends the family into document understanding. The lightweight model extracts Arabic text and structured content from images and documents, including tables, mathematical formulas and different document sections, with potential uses in document digitisation, searchable archives and automated information extraction.

“Our language belongs in the future we are building. With Falcon-Emirati, we are putting our knowledge and expertise behind that belief. There is real pride in building technology that understands our people. And there is real freedom in having the capability here in the UAE to develop it further, set our own priorities and turn our ambitions into something people can use,” Faisal Al Bannai, Advisor to the UAE President for Strategic Research and Advanced Technology Affairs, and Secretary-General of ATRC said in a statement.

Advertisement

“Sovereign AI capability must reflect the language used in daily life. Emirati Arabic carries distinctive expressions, cultural references and ways of communicating. Falcon-Emirati and Falcon-ASR will help ensure that the next generation of AI understands not only Arabic but how Emirati communities actually speak it,” Dr. Najwa Aaraj, Chief Executive Officer of TII added.

“These capabilities empower developers, researchers and communities across the region to build on Emirati language and culture, while expanding what is possible for Arabic AI more broadly. Across Falcon-Emirati, Falcon-ASR and Falcon-OCR-Arabic, we have applied targeted specialisation to language, speech and document understanding, addressing areas where Arabic AI still has significant room to advance,” Dr. Hakim Hacid, Chief Researcher of TII’s Artificial Intelligence and Digital Research Centre further explained.

More News