Current DysVoxa beta demonstrations
These videos document the experimental Windows workflow: setup, direct text-to-speech, short-phrase recording, local recognition, reviewable alternatives, and user-triggered Piper output. They demonstrate specific runs, not validated performance across speakers.
What these videos do - and do not - show
Current interface
The recordings show a Windows beta build. Packaging, labels, candidate handling, and other details may change.
Review-first workflow
They show phrase recording, local recognition, text review, and an explicit Speak action rather than automatic speech.
Not validation
A successful clip does not establish reliability for another speaker, severity, phrase, computer, or environment.
1. First launch
A walkthrough of the current first-run setup. Interfaces and packaging may change while the experimental Windows beta is strengthened.
2. Generating speech from text
Direct text-to-speech does not use or clone the speaker's recorded voice. VB-CABLE routing is experimental and should be tested with a remote partner before use in a call.
3. Talking in standard and advanced mode
This demonstration compares Standard (Parakeet) and Advanced (Whisper) recognition. They may interpret a phrase differently, and neither mode nor any correction suggestion is guaranteed to recover the intended meaning.
Follow development or join testing
Register interest in the Windows beta, review its current limitations, or support the testing and engineering work.
Fund dysarthria speech recognition that stays free to use.
One-time donations, Indiegogo backing, and organizational sponsorship pay for safety work, Windows packaging, and broader evaluation. SIA DysVoxa is a registered company in Latvia, so gifts may not be tax-deductible.