Hugging Face introduces Falcon-ASR, a 1.6B parameter speech recognition model for Arabic

Falcon-ASR, a 1.6 billion parameter speech recognition model developed at the Technology Innovation Institute (TII) in Abu Dhabi, supports Arabic (including Emirati dialect), English, French, Spanish, and Portuguese. Across six Arabic test sets, it achieved an average Word Error Rate (WER) of 20.92%, surpassing the best published result of 23.17%. In an internal Emirati evaluation, it recorded a WER of 22.73% and a Character Error Rate (CER) of 10.19%, achieving the lowest rates among compared systems, 4.07 percentage points lower than Qwen3-Omni.