Voice biomarker research
CortexVoice treats speech as a research signal.
CortexVoice looks at how recorded speech, short questionnaires and analysis models can support studies of cognitive and emotional change, without overstating what the results mean.
Overview
Speech carries patterns worth measuring carefully.
Vocal features such as pitch movement, pauses and articulation shift with fatigue, mood and cognition. CortexVoice studies those features as research data, never as a diagnosis or medical advice.
Participants complete guided voice tasks and short surveys. Inside Cortex Dao, the project sits next to NeuroBand and CortexOS as a low-effort way to contribute consented signals.
In progress
Voice features
- Pitch movement
- Vocal tremor
- Speaking rate
- Pause length
- Articulation
- Prosody and volume
- Base frequency
- Jitter and shimmer
- Harmonic-to-noise ratio
- Onset timing
- Voice quality
- Intelligibility
Project functions
01
Recording module
Collects voice tasks and related answers in a structured, repeatable format.
02
Analysis layer
Turns raw audio into feature tables that researchers can compare across sessions.
03
Participant controls
Every recording stays tied to a permission its owner can withdraw.
04
Privacy by design
Studies work on de-identified features, not on open access to raw audio.
05
Study builder
Researchers set up scoped projects around datasets that participants approved.
06
Network integration
Feeds voice research into CortexOS, community compute and governance.
Why it matters
Low-friction signals widen who can take part.
Talking into a phone is easier than any lab visit. CortexVoice gives the network a research lane built on that simplicity, with careful language about validation and privacy.