--- title: Introducing Falcon ASR | daily.dev description: TII introduces Falcon-ASR, a 1.6B parameter speech recognition model focused on Arabic with particular strength in the Emirati dialect, also supporting... canonical: https://daily.dev/posts/introducing-falcon-asr-jh3dtlnlp twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Introducing Falcon ASR | daily.dev og:description: TII introduces Falcon-ASR, a 1.6B parameter speech recognition model focused on Arabic with particular strength in the Emirati dialect, also supporting... og:url: https://daily.dev/posts/introducing-falcon-asr-jh3dtlnlp og:image: https://api.daily.dev/og/posts/jH3DTLNLp.png og:image:alt: Introducing Falcon ASR og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Introducing Falcon ASR **[Hugging Face](https://daily.dev/sources/huggingface)** · 7 min read · 1 upvotes · 0 comments ## Summary TII introduces Falcon-ASR, a 1.6B parameter speech recognition model focused on Arabic with particular strength in the Emirati dialect, also supporting English, French, Spanish, and Portuguese. On the Open Universal Arabic ASR Leaderboard protocol, it achieves 20.92% average WER across six Arabic test sets, beating the best published result of 23.17%. On an internal Emirati evaluation it scores 22.73% WER, ahead of Qwen3-Omni-30B at 26.80%. It also supports word-level timestamps and achieves 5.74% mean WER on English benchmarks. The model is available to try via a Hugging Face demo space, with API access and native apps planned. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What word error rate does Falcon-ASR achieve on Arabic speech recognition benchmarks? Falcon-ASR achieves an average word error rate of 20.92% across six Arabic test sets on the Open Universal Arabic ASR Leaderboard protocol, compared with 23.17% for the next best published result, Audar-ASR-V1-Turbo (2.35B parameters). Falcon-ASR has 1.6 billion parameters and also records an average character error rate of 8.79%. _Teams comparing Arabic ASR models can track benchmark releases like this one on daily.dev._ ### How well does Falcon-ASR handle the Emirati Arabic dialect compared to other speech models? Falcon-ASR records the lowest word and character error rates among compared systems on an internal Emirati evaluation, scoring 22.73% WER and 10.19% CER. That is 4.07 percentage points better than the next best model, Qwen3-Omni-30B-A3B-Instruct, which scored 26.80% WER, and ahead of Cohere Transcribe Arabic and Audar-ASR-V1-Turbo. _Developers building dialect-aware speech tools can follow comparisons like this on daily.dev._ ### Which languages does the Falcon-ASR model support besides Arabic? Falcon-ASR supports English, French, Spanish, and Portuguese in addition to Arabic, using the same model weights across all five languages without requiring a language flag. On English, it achieves a mean word error rate of 5.74% across seven public test sets from the Hugging Face Open ASR Leaderboard, including 1.75% on LibriSpeech clean and 11.86% on Earnings-22. _Those evaluating multilingual ASR options can keep up with model releases like this via daily.dev._ ## Similar posts on daily.dev - [How to Fine-Tune Nemotron 3.5 ASR for Your Language, Domain, or Accent](https://daily.dev/posts/how-to-fine-tune-nemotron-3-5-asr-for-your-language-domain-or-accent-lmayxxp44) · Hugging Face · 0 upvotes · 0 comments --- Tags: [#speech-recognition](https://daily.dev/tags/speech-recognition), [#huggingface](https://daily.dev/tags/huggingface) [View this post on daily.dev](https://daily.dev/posts/introducing-falcon-asr-jh3dtlnlp) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Introducing Falcon ASR","url":"https://daily.dev/posts/introducing-falcon-asr-jh3dtlnlp","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/introducing-falcon-asr-jh3dtlnlp"},"datePublished":"2026-10-07T13:23:52.775Z","dateModified":"2026-10-07T13:24:23.952Z","description":"TII introduces Falcon-ASR, a 1.6B parameter speech recognition model focused on Arabic with particular strength in the Emirati dialect, also supporting...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/9b031ba927ea688aeb96d50d0d7ca10e?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/9b031ba927ea688aeb96d50d0d7ca10e?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Hugging Face","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Hugging Face","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/f1f55c67d81a4330acf5b90b26b0c8e1","url":"https://daily.dev/sources/huggingface"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/introducing-falcon-asr-jh3dtlnlp","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"speech-recognition,huggingface","timeRequired":"PT7M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Hugging Face","item":"https://daily.dev/sources/huggingface"},{"@type":"ListItem","position":3,"name":"Introducing Falcon ASR"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/introducing-falcon-asr-jh3dtlnlp#faq","mainEntity":[{"@type":"Question","name":"What word error rate does Falcon-ASR achieve on Arabic speech recognition benchmarks?","acceptedAnswer":{"@type":"Answer","text":"Falcon-ASR achieves an average word error rate of 20.92% across six Arabic test sets on the Open Universal Arabic ASR Leaderboard protocol, compared with 23.17% for the next best published result, Audar-ASR-V1-Turbo (2.35B parameters). Falcon-ASR has 1.6 billion parameters and also records an average character error rate of 8.79%. Teams comparing Arabic ASR models can track benchmark releases like this one on daily.dev."}},{"@type":"Question","name":"How well does Falcon-ASR handle the Emirati Arabic dialect compared to other speech models?","acceptedAnswer":{"@type":"Answer","text":"Falcon-ASR records the lowest word and character error rates among compared systems on an internal Emirati evaluation, scoring 22.73% WER and 10.19% CER. That is 4.07 percentage points better than the next best model, Qwen3-Omni-30B-A3B-Instruct, which scored 26.80% WER, and ahead of Cohere Transcribe Arabic and Audar-ASR-V1-Turbo. Developers building dialect-aware speech tools can follow comparisons like this on daily.dev."}},{"@type":"Question","name":"Which languages does the Falcon-ASR model support besides Arabic?","acceptedAnswer":{"@type":"Answer","text":"Falcon-ASR supports English, French, Spanish, and Portuguese in addition to Arabic, using the same model weights across all five languages without requiring a language flag. On English, it achieves a mean word error rate of 5.74% across seven public test sets from the Hugging Face Open ASR Leaderboard, including 1.75% on LibriSpeech clean and 11.86% on Earnings-22. Those evaluating multilingual ASR options can keep up with model releases like this via daily.dev."}}]} ```