Speech biomarkers mapped to the clinical scales your trial already runs.
New recordings on the participant's phone, tablet or computer, or recordings your study already holds, read on the clinical scale the study uses.
Speech is a window into the nervous system
Speaking draws on the brain, the breath, the larynx and the articulators. One short recording carries signal from all of them.
What the scales record at visits, read from speech between them
Voice is inherently multi-modal: a couple of minutes of speech carries cognitive and motor signal at once. We decompose it into interpretable, disease-specific domains.
Mapped to the trial's scale
A digital measure that scales.
Non-invasive
A short recorded task. No imaging, no clinic burden on the participant.
The participant speaks for a couple of minutes into a phone or browser. Nothing is worn, nothing is implanted, nothing is scanned. Because the burden is low, the task repeats as often as the protocol needs.
No dedicated device
Runs on the participant’s own phone, tablet or computer. Nothing to ship.
There is no hardware to ship, provision, charge, track or recover at the end of the study.
Cross-language
One architecture across languages.
Our speech dataset is collected in several languages against one shared protocol, with locally appropriate stimuli and metadata aligned across the corpus. Fixed stimuli are what make cross-language comparison possible: without them, two recordings differ both because the speakers differ and because the elicitation did.
Low burden
Low participant effort supports frequent, remote sampling between site visits.
A couple of minutes at home, on a schedule the protocol sets, needs no travel and no site time. That is what makes weekly or higher-cadence sampling routine, and frequent sampling is the whole point of a between-visit measure.
Objective and repeatable
Quantitative features, scored the same way at every visit.
Self-administered tasks present the same stimulus at every visit, and the output is a set of named features.
Extends reach
Automated analysis extends assessment to settings where frequent in-clinic testing is impractical.
Assessment reaches participants who live far from a site, cannot travel easily, or are enrolled somewhere without the staff to administer a scored cognitive test. The constraint stops being clinic capacity and becomes whether the person can speak for a couple of minutes.