The Science

The science behind the voice.

Voice is a continuous, involuntary readout of your autonomic nervous system, respiratory mechanics and neurological state. Biota1 listens to what your body is actually saying.

Premise

Why voice?

Every time you speak, your vocal cords vibrate under the influence of vagal tone, diaphragm mechanics, laryngeal muscle tension, and cortical arousal. These forces leave measurable signatures in your acoustic output. Micro-variations in pitch, harmonic structure, breathing patterns, timing.

No sensor placed on the skin can access the biomechanics of phonation. Voice is the only non-invasive window into these systems.

10ms
Vagal response latency
48kHz
Sampling rate
48+
Acoustic features
30s
Speech needed

The Pipeline

From sound
to signal.

Tap any stage below to follow how 30 seconds of speech becomes 48+ wellness biomarkers, without ever leaving the device.

Signal Stream · Active
Biota1 Engine v.2.04

Anatomy

A direct line
to your nervous system.

The vagus nerve innervates the larynx directly. State changes propagate through phonation in milliseconds.

01020304

Anatomical map

The vagal pathway.

Physiological readings · live

Your body,
read from a sentence.

Thirty seconds of speech produces a full physiological snapshot — heart rate, HRV, respiratory mechanics, cortisol rhythm, cognitive load, hydration. No cuff. No electrode. No skin contact.

Heart rate · resting

68 BPM

Inferred from vocal micro-tremor & subglottal pressure

60708090
82
Recovery
Composite ANS index
71
Energy
Vocal vigor score

HRV proxy · 14-day

68 ms RMSSD

Respiratory sinus arrhythmia, acoustically derived

030609014 DAYS · RMSSD (ms)

Respiratory

14/min

Cognitive load

34%

Low — disfluency rate within baseline

Cortisol proxy · diurnal

Healthy curve

Cortisol awakening response detected · normal evening decline

06:0010:0014:0018:0022:00

Hydration

76%

Mucosal resonance · vocal fold lubrication

SAM suggests +200ml across afternoon.

Acoustic feature map

48+ features per check-in.

MFCC · jitter · shimmer · F0 · formants · HNR · prosody

48kHz
Sample rate
25ms
Frame size
10ms
Hop length
8kHz4kHz2kHz1kHz500Hz250Hz30s · MFCC · F0 · FORMANTS · HNR

48+ measured

What we measure.

Eight wellness domains. Over fifty-two acoustic and derived features per check-in. Click any row.

Core acoustic measures of phonation quality and vocal fold stability.

Example markers. Fundamental frequency (F0), jitter, shimmer, harmonics-to-noise ratio

Validation

AUC by domain

0.899
Mean AUC · 46 biomarkers
Stress
.92
Sleep
.88
Respiratory
.94
HRV Proxy
.86
Cognitive
.89
Energy
.87
Prosody
.91
Hydration
.84

Peer-reviewed study submitted to Frontiers in Digital Health (Manuscript ID 1813478). Co-authored by Dr Adrian Attard Trevisan, Andrea Sprio, and Prof. Frederick R. Carrick.

Architecture

Privacy is not a feature.
It is the architecture.

01

On-Device Only

All AI inference runs on the embedded edge processor. Audio never leaves the device.

02

No Audio Upload

BLE transmission contains only numerical biomarker values. Voice is discarded on-chip.

03

GDPR by Design

Architected from the ground up to be GDPR-compliant. No personal audio data to breach.