Falcon ASR: 新しいアラビア語向け音声認識モデル

WER هو معدل خطأ الكلمات

it also supports English, French, Spanish and Portuguese.

CER هو معدل خطأ الأحرف

حقق Falcon-ASR متوسط معدل خطأ في الكلمات بلغ 20.92% عبر ست مجموعات اختبار باللغة العربية

نموذجنا للتعرف على الكلام العربي بحجم 1.6 مليار معلمة

speech recognition model for Arabic

Falcon-ASR has the lowest WER and CER among the systems compared here.

our 1.6 billion parameter speech recognition model for Arabic

وهو يدعم أيضًا اللغات الإنجليزية والفرنسية والإسبانية والبرتغالية

across six Arabic test sets

في معهد الابتكار التكنولوجي (TII) في أبوظبي

Falcon-ASR achieved an average word error rate of 20.92% across six Arabic test sets

Emirati WER (TII evaluation): 22.73%

Its WER is 4.07 percentage points below Qwen3-Omni, the next best result.

Lower values are better for both metrics.

طوّرنا النموذج في معهد الابتكار التكنولوجي (TII) في أبوظبي

In our internal Emirati evaluation, Falcon-ASR achieved 22.73% WER and 10.19% CER:

In our evaluation, Falcon-ASR achieved an average word error rate of 20.92% across six Arabic test sets, compared with the best published result of 23.17% in the leaderboard snapshot we used. On our internal Emirati evaluation, it recorded the lowest word and character error rat…

We’re introducing Falcon-ASR, our 1.6 billion parameter speech recognition model for Arabic, with a particular focus on the Emirati dialect. Developed at the Technology Innovation Institute (TII) in Abu Dhabi, it also supports English, French, Spanish and Portuguese.