Accessibility
Who Rubil is built for, and what we haven't tested.
This page says who Rubil is built for and what we haven't tested yet. It also explains how we test.
Who it's for
People who can speak comfortably and find typing hard.
Typing is painful or physically hard
People managing repetitive strain injury, chronic pain, arthritis, or limited hand or arm movement, and people who use a computer one-handed.
Getting words typed is the hard part
People who find writing slow or overwhelming, including some people with ADHD, dyslexia or dysgraphia.
One caution if you rely on Rubil without rereading: speech recognition can mishear a word, and the formatted text can still look right. If proofreading is hard for you, keep that in mind for anything important.
Status
What we've tested. What we haven't.
- Built for this
- Typing hurts
- Your hands are limited
- Writing feels slow or overwhelming
- Testing next
- Screens and reading wear you outTremor, MS, fatigue conditions, brain injury, dyspraxia
- Not tested yet
- You can't use your hands, or can't see the screenVoice Control, head or mouth pointers, VoiceOver, screen magnifiers
- Your speech is affectedSpeech affected by Parkinson's, cerebral palsy, ALS or a stroke
- Rubil in Chrome
Where Rubil has not been tested: speech
For stuttering, Rubil takes out repeated sounds and words, and a pause doesn't end the take.
If your speech is affected by dysarthria, for example from Parkinson's, cerebral palsy, ALS or stroke, we have not tested Rubil with you, and we won't claim it works.
Published research on the class of speech recognition Rubil uses, not on Rubil itself, reports steep drops in accuracy with these speech patterns:
- 3.4% to 36.3%word error rate, typical speech against Parkinson's-related dysarthria, in a study of 211 speakers.
- 12.7% to 21.5%average word error rate on sentences, in Whisper tests on standard dysarthric speech collections.
- 3.1% to 19.8%word error rate, fluent against disfluent speech, in one controlled study.
Every figure is linked under Sources. We'd rather give you the research than a reassurance.
Why we won't just say "give it a try"
Research on recognizers for dysarthric speech found that Whisper often invents words, producing transcripts that read smoothly and are wrong. If a smooth but wrong transcript reaches Rubil's formatting step, the result is clean, confident text you never said. For anyone who can't easily proofread, that's worse than a tool that visibly fails.
What would change this
Before we claim anything for dysarthria, we plan to run a small trial with volunteers who have mild dysarthria, supervised by a speech-language pathologist and based on consent. We'll publish the result either way, including the limits. We haven't set a date.
If you'd like to help test, whether you stutter or your speech is affected in another way, we'd be grateful. Write to [email protected]. We'll tell you exactly what testing involves before you agree to anything, and you can stop at any point.
Screen readers
We haven't tested Rubil with VoiceOver or any other screen reader yet. We won't claim screen reader support until blind testers can finish the main tasks on their own.
How we test
- Keyboard only. With the mouse and trackpad unplugged, we go through the whole journey: setup, dictation, recovering a failed paste, History, every setting, and quitting.
- Automated checks. We scan the Rubil window, the recording indicator and setup with an automated accessibility checker.
- VoiceOver walkthrough. We log every step that is silent or can't be reached. This is to find problems, and it supports no claim.
- People who find typing hard. A small group of volunteers uses Rubil for two weeks and tells us what breaks.
Formal conformance
Rubil has not been through a formal WCAG audit, and there is no VPAT for it. If you need either for a procurement or institutional review, write to us and we'll tell you where it stands.
Sources
None of these is a study of Rubil. They describe the class of speech recognition Rubil uses.
- Speech Accessibility Project, Parkinson's dysarthria study (University of Illinois, JSLHR)
- Fine-Tuning ASR for People with Parkinson's (arXiv 2409.19818)
- Bridging ASR and LLMs for Dysarthric Speech (arXiv 2508.08027)
Ready to stop typing?
Try Pro free for 7 days. Then 1,000 words a week, free forever. No credit card.