The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
TII’s Falcon-ASR is a 1.6-billion-parameter speech-to-text model that reports a 20.92% average word error rate across six Arabic test sets. That is an aggregate result under the Open Universal Arabic ASR Leaderboard protocol—not a promise that any individual recording will have that error rate. TII also reports a separate 22.73% WER on its internal Emirati evaluation, which should not be treated as the same test.
What is Falcon-ASR?
Falcon-ASR is an automatic speech recognition model developed by the Technology Innovation Institute (TII). It converts spoken audio into text and has 1.6 billion parameters. TII lists Emirati Arabic, Modern Standard Arabic (MSA), English, French, Spanish, and Portuguese as supported languages. Its project article says the same model weights handle these language groupings without requiring a language flag. TII’s project article and its Falcon model overview describe the model and release.
TII says its training included Emirati Arabic, MSA, other Gulf and Arabic dialects, and English. The institute also describes training augmentation and evaluation conditions involving background noise, overlapping speech, music, room reverberation, telephone effects, and changes to speed and pitch. Those are developer-reported methods, not a guarantee that the model will perform consistently on every noisy, accented, or overlapping recording.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
How accurate is Falcon-ASR for Arabic speech?
Public six-test-set Arabic result
TII reports an average of 20.92% WER and 8.79% character error rate (CER) across six Arabic test sets, following the Open Universal Arabic ASR Leaderboard protocol. The protocol’s average gives the six sets equal weight; lower WER and CER mean fewer differences from the reference transcripts. These are TII-reported results in its evaluation tables.
WER counts word substitutions, deletions, and insertions against a reference transcript. CER makes a similar comparison at the character level. The percentages summarize performance over the evaluated test data; they do not mean every recording has that proportion of errors, nor do they reveal how a particular error affects meaning. TII’s article does not provide a confidence interval or a consumer-facing interpretation of the score, so the aggregate should be read as a benchmark result rather than a prediction for your own audio.
How the comparison stands in TII’s snapshot
In the comparison table TII says it checked on 30 September 2026, the strongest listed published comparator was Audar-ASR-V1-Turbo at 23.17% average WER and 9.23% CER. Falcon-ASR’s listed WER is 2.25 percentage points lower. The other models shown are included below; these are values from TII’s stated snapshot, not an independently verified live ranking.
| Model | Average WER | Average CER | Evidence and scope |
|---|---|---|---|
| Falcon-ASR | 20.92% | 8.79% | TII-reported average over six Arabic test sets using the Open Universal Arabic ASR Leaderboard protocol |
| Audar-ASR-V1-Turbo | 23.17% | 9.23% | Published comparator in TII’s 30 September 2026 snapshot |
| Cohere Transcribe Arabic (07-2026) | 25.87% | 11.80% | Published comparator in TII’s 30 September 2026 snapshot |
| omniASR LLM 7B | 28.32% | 12.52% | Published comparator in TII’s 30 September 2026 snapshot |
The table supports a narrow conclusion: on the six-set protocol and comparison snapshot TII presents, Falcon-ASR has the lowest listed WER and CER among these entries. It does not establish performance on every Arabic dialect, recording condition, or deployment workload.
Separate internal Emirati evaluation
TII separately reports 22.73% WER and 10.19% CER on an internal Emirati evaluation using additional Emirati and Gulf speech, held-out recordings, and human-validated transcripts. TII lists Qwen3-Omni-30B-A3B-Instruct at 26.80% WER and 12.72% CER as the next-lowest WER shown in that internal table. This is TII’s own evaluation, distinct from the public six-test-set Arabic average; the figures should not be combined or compared as though they came from the same test. Details appear in the project article and launch announcement.
Rank #3
English result is a different benchmark
TII also reports a 5.74% mean WER across seven public English test sets. That is a separate English evaluation, not an Arabic score; the per-set figures TII lists range from 1.75% on LibriSpeech clean to 11.86% on Earnings-22.
Does Falcon-ASR support Emirati Arabic?
Yes. TII explicitly lists Emirati Arabic among the model’s language varieties and gives it special attention. Its separate internal evaluation is relevant evidence, but it is TII-reported and does not establish identical accuracy across all Emirati speakers, topics, accents, or recording conditions. For a real deployment, assess recordings representative of your speakers and use case rather than relying only on an overall benchmark.
What can Falcon-ASR be used for?
TII identifies speech transcription, subtitling, searchable recordings, meeting and interview transcription, accessibility tools, and voice services as possible applications. It also states that word-level timestamps are a capability. These are described use cases and features; the release materials do not establish a specific production service or guarantee suitability for a particular workflow.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →How can I try Falcon-ASR?
TII points readers to a Hugging Face demo where they can try recordings. The same project article says API access and native applications are planned, so those should not be assumed to be generally available now. The materials do not specify exact hardware requirements or Falcon-ASR-specific checkpoint licensing and commercial terms. Check the current model documentation and terms before integrating it into a product.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
