Introducing Falcon ASR

Hugging Face and the Technology Innovation Institute (TII) in Abu Dhabi have introduced Falcon-ASR, a 1.6 billion parameter speech recognition model tailored for Arabic with a primary focus on the Emirati dialect. In addition to Arabic, the model supports English, French, Spanish, and Portuguese, utilizing the same underlying weights across all five languages without requiring a language flag. According to evaluations cited by the developers, Falcon-ASR achieved an average word error rate (WER) of 20.92% across six Arabic test sets, outperforming previous published leaderboard results. In an internal Emirati evaluation using held-out recordings and human-validated transcripts, Falcon-ASR recorded a WER of 22.73% and a character error rate (CER) of 10.19%, outperforming compared systems such as Qwen3-Omni. For English, the model achieved a mean WER of 5.74% across seven public test sets from the Hugging Face Open ASR Leaderboard. Falcon-ASR also supports word-level timestamps, linking transcribed words to their precise positions in audio files. The training data incorporated various acoustic challenges, including background noise, overlapping speech, music, room reverberation, and telephony effects, to prepare the model for everyday recordings and meetings. The developers built Falcon-ASR upon prior Falcon3-Audio work and have made a demo space available publicly for user testing, with API access and native applications planned for the future.