Free
DysVoxa Free
- Guided first-run setup
- Live speech recognition
- Local and optional cloud correction
- Output routing to speakers or virtual microphone
- Quick phrases
Clear speech, on Windows
AI that helps people with dysarthria be understood again
DysVoxa is a Windows desktop app for people with dysarthria. Speak into your microphone; the app listens, understands your words, optionally cleans them up with AI, and speaks them aloud through your speakers or a virtual microphone for calls and meetings. Everything runs locally on your machine. Built by a founder living with spinocerebellar ataxia and ataxic dysarthria, who also has limited mobility.
The problem
46 million people worldwide live with dysarthria. Their thoughts are clear, but their speech muscles don't cooperate. Words come out slurred, slow, or hard to understand. Conditions like stroke, Parkinson's, ALS, cerebral palsy, and ataxia can all cause it.
The real cost goes beyond unclear audio. After enough failed conversations — repeating yourself, being misheard, having someone else speak for you — many people start avoiding calls, group settings, and speaking only when necessary. Communication breakdown leads to isolation.
Why I built this
I built DysVoxa because I needed it myself. After being diagnosed with spinocerebellar ataxia, everyday communication started carrying friction: repeating myself on calls, losing momentum in conversations, watching routine interactions become tiring. My mobility is now significantly limited — which makes reliable remote communication matter even more.
As a technologist, I approached that as an engineering problem: reduce the communication breakdown, measure what works, and keep iterating until the system becomes useful in daily life. That's what DysVoxa is.
— Igor, Founder
How it works
1. Listen
Detects when you are speaking and captures your voice.
2. Understand
Converts speech to text with engines tuned for dysarthric speech.
3. Correct
Optionally improves wording using on-device AI, or cloud correction when you opt in.
4. Speak
Reads the result aloud with natural text-to-speech, routed to speakers, a virtual microphone, or both.
What makes DysVoxa different
Built for impaired speech
Not repurposed dictation software. Designed from day one for speech that standard tools fail to recognize.
Privacy-first
Speech recognition and voice output run entirely on your computer. Biometric voice data and audio never leave your device. In the current beta, there is no telemetry or behavior tracking.
Guided first-run setup
Lists what is installed, what is missing, and downloads required components with progress, cancel, and retry. No command line, no manual file copying.
One application, no setup complexity
A single desktop application with everything built in. No servers, no extra software, no technical expertise needed.
Uses your natural speech
Not a symbol board or pre-programmed phrases. You speak naturally and DysVoxa makes it clearer.
Affordable alternative
Planned to cost a fraction of dedicated AAC hardware, while still being a supported, maintained product.
Free and Pro
DysVoxa Free is built for daily use without a subscription. DysVoxa Pro adds personalized speech recognition trained to your voice.
Pro
DysVoxa Pro
- Everything in Free
- Personalized speech recognition
- Guided voice recording with SLP-style workflow
- Train a voice adapter on your CPU
- Adapter management and roll-back to baseline
- Tuning progress and before/after quality reports
This model keeps entry simple for users while creating a credible path to long-term sustainability.
How funding and revenue fit together
Now: bridge from beta to trusted product
Donations and grants fund user testing, legal compliance, accessibility work, and day-to-day operations before recurring product revenue is mature.
Later: paid plans sustain long-term support
Freemium access broadens adoption, while paid personal, clinical, and partner plans fund maintenance, support, and product improvement over time.
Why back this now
46M+
People worldwide living with dysarthria
Working beta today
Core speech loop already runs in real time
Privacy-first by design
Biometric voice features never leave the device
Where we are today
DysVoxa is a working Windows desktop app. The core loop — listen, understand, correct, speak — runs in real time on the founder's machine. It has been tested with ataxic dysarthria.
What works now
- Full speech pipeline: listen, understand, correct, speak
- Guided first-run setup — no command line, no manual file copying
- Local speech recognition and text-to-speech
- Optional cloud correction (recognized text only, never audio)
- Output routing to speakers or virtual microphone for calls
- Quick phrases
What's next
- Testing with people who have speech impairments
- Measuring real-world effect and usability
- GDPR compliance and accessibility audit
- Public launch as a paid, supported product
The technology works. What's missing is the expensive bridge between a working prototype and a trustworthy product: user testing, usability validation, legal compliance, and operational support. That's what support and early funding make possible.
Why this matters beyond technology
Research on dysarthria consistently shows not just reduced speech clarity, but reduced communicative participation. People withdraw from conversations, from calls, from the social fabric of daily life. DysVoxa is not therapy and not a cure. But if it helps someone stay in one more phone call, one more appointment, one more conversation they would otherwise avoid — that's the point.
Help make this real
Your donation funds user testing, usability research, and the operational work needed to turn a working beta into a product people can trust and rely on.
Privacy
- Biometric voice data and audio never leave your device. Speech recognition and synthesis run locally on your device.
- No account is required for the free default profile (bring your own API key).
- The current beta includes no telemetry or behavior tracking.
- Optional text-only cloud correction sends recognized text (never audio) and can be fully disabled.
- Designed with GDPR principles: privacy by design, data minimisation, purpose limitation.
Clinic and sponsor partnerships
If you represent a clinic, school, AAC team, foundation, or grant program, DysVoxa can provide pilot onboarding and a simple implementation plan. All partnership tracks preserve the same rule: biometric voice data and audio never leave your device.
Honest limits
- Assistive software, not a medical device. Not a cure or a replacement for professional care.
- Built first for dysarthria. Broader motor-speech validation is still in progress.
- Windows desktop. No mobile app yet.
Contact