TII Releases Falcon-ASR: 1.6B Arabic Speech Model Hits 20.92% WER

Abu Dhabi's Technology Innovation Institute (TII) released Falcon-ASR, a 1.6B-parameter Arabic speech recognition model focused on the Emirati dialect. It posts a 20.92% average word error rate across 6 Arabic test sets, 2.25 points better than the best published entry in TII's leaderboard snapshot. No weights or API are available yet — only a Hugging Face demo.

• 1.6B parameters, 20.92% average WER across 6 Arabic test sets
• 2.25 points better than Audar-ASR-V1-Turbo (23.17%)
• One set of weights covers 5 languages: Arabic, English, French, Spanish, Portuguese
• No weights, license or API yet — only a 60-second demo