Contact Us
Voice Biomarker · Neurology

Hearing Concussion
Before It's Too Late

A prospective study in partnership with Pharnaceutical Sponsor — applying Sona-2's Large Acoustic Model to real-world TBI detection in contact sport athletes. What started as video recordings from a rugby pitch in Australia is now a formal 18-week longitudinal study.

Concussion leaves a fingerprint in the voice. Disrupted phonation, altered prosody, spectral texture changes — signals too subtle for a sideline assessment but within reach of an acoustic model trained on the right data. This is that study.

ModalityVoice / Acoustic AI
ConditionTBI / Concussion
PhaseProspective Study
StatusActive
Sona-2 TBI Output · Sample Athletes Pilot
A
Athlete A
Rugby · Confirmed concussion prior season
90.4%
TBI Positive
B
Athlete B
Rugby · Confirmed concussion prior season
78.4%
TBI Positive
Samples collected weeks post-injury, no acute baseline. Prospective study now underway.
Pharnaceutical Sponsor Prospective Protocol Active
30–50 confirmed concussed athletes. Blood biomarkers across acute and sub-acute phases. SCAT6 comparator. Voice audio collected as additional modality. Longitudinal multi-modal capture.
01Origin

How This Partnership
Came Together

Three independent vectors — a biomarker platform, clinical trial infrastructure, and elite sport access — converged into a single prospective study.

The TBI program started with a question a clinical collaborator brought to Amplifier: can voice detect the residue of a brain injury that no imaging can see? The neuromechanics are real. Concussion disrupts phonation at the brainstem level, altering prosody, timing, and spectral texture in ways humans cannot perceive but acoustic models can.

"The brain injury shows up in the voice before it shows up in the clinic. We just needed a model that could read it."

The clinical team connected Amplifier with a trial infrastructure partner in Australia — purpose-built for contact sport research with direct access to elite athletes and established biomarker protocols. The trial partner had the athletes, the longitudinal blood panel design, and the methodology. Amplifier had Sona-2. The fit was immediate.

To test the hypothesis before committing to a full prospective design, the trial team sent video recordings of athletes confirmed concussed the prior season, recorded weeks post-injury. The Amplifier AI team ran the audio through Sona v2.1.0. One athlete returned 90.4% TBI positive. The other returned 78.4%. The trial team flagged the limitations themselves — post-acute, no baseline. But the signal direction was right. That was enough. The prospective study is now underway.

Clinical Lead
Originated the partnership, structured the collaboration framework, managing sport organization relationships and the Australia study.
Trial Infrastructure Partner · Australia
Purpose-built clinical trial operation. Elite contact sport athlete access, longitudinal blood biomarker protocols, SCAT6 methodology. Running the prospective study on the ground.
Amplifier Health
Built the Sona-2 Large Acoustic Model platform. Driving global sports licensing commercial strategy.
02Pilot Signal

Sample Athletes —
The Signal That Started Everything

Confirmed concussions. One model. The results weren't definitive — they were directionally decisive.

Athlete A
Rugby · Male · Elite
Post-Acute
ModelSona v2.1.0
Sample TypeVideo → Stripped Audio
Post-Injury TimingWeeks post-injury
Acute BaselineNone
TBI Confidence90.4%
Signal persisted weeks post-injury with no baseline comparison. Direction confirmed.
Athlete B
Rugby · Female · Elite
Post-Acute
ModelSona v2.1.0
Sample TypeVideo → Stripped Audio
Post-Injury TimingWeeks post-injury
Acute BaselineNone
TBI Confidence78.4%
Lower confidence consistent with post-acute attenuation and individual variation.
What This Run Was
A directional feasibility test. The trial team sourced video recordings from athletes confirmed concussed the prior season. The Amplifier AI team stripped the audio and ran it through Sona v2.1.0 — no preprocessing adjustments, no baseline calibration, no acute reference point. Raw inference on post-acute voice samples.
What This Run Was Not
A formal validation study. Samples were captured weeks post-injury with no pre-injury baseline. This phase was designed to establish signal direction, not sensitivity or specificity.
Why It Still Mattered
Every sample returned TBI-positive above threshold. In the wrong direction, that result kills the hypothesis. In the right direction, it justifies the cost of a real study. It pointed right.
03Study Design

The Real Study.
Now Underway.

Longitudinal. Acute-phase capture. Multi-modal biomarker protocol. This is where the validation data gets built.

Pre-Study
Protocol finalization, voice modality integration, athlete enrollment. Pharnaceutical Sponsor infrastructure confirmed.
Acute Phase
Confirmed concussion triggers immediate voice capture alongside blood biomarker draws. First longitudinal comparison points established.
Sub-Acute Phase
Continued voice and blood biomarker collection. SCAT6 symptom evaluations running in parallel. Recovery trajectory tracked.
End of Study
Full dataset assembled. Voice biomarker performance analyzed against blood biomarkers and SCAT6. Commercial deployment readiness assessed.
PopulationConfirmed concussed elite contact sport athletes
Duration18 weeks
ComparatorSCAT6
Blood BiomarkersMultiple draws, acute & sub-acute phases
Voice ModalityAdditional modality alongside blood panel
IRBNot required for this phase
LocationAustralia · Pharnaceutical Sponsor
ModelSona v2.1.0
04Clinical Framework

Detect. Protect. Return.

Three words. One clinical pathway. The framework that takes voice biomarker data from sideline to return-to-play decision.

01
Detect
Sideline Identification
Current sideline tools rely on self-report, symptom checklists, and gross cognitive testing. Athletes hide symptoms. Assessments miss subclinical injury. Sona-2 adds an objective acoustic layer — a passive, non-invasive signal that doesn't depend on athlete cooperation or clinician judgment.
Sideline AssessmentPassive CaptureObjective Signal
02
Protect
Recovery Monitoring
Without objective longitudinal tracking, return-to-play decisions rest on symptom resolution alone — which normalizes before the brain does. Sona-2 tracks acoustic recovery across the sub-acute window, giving medical staff a continuous biomarker signal to inform hold decisions.
Longitudinal TrackingRecovery MonitoringRisk Reduction
03
Return
Clearance Decision
Athletes, teams, and medical staff need a defensible, data-backed clearance signal. Sona-2's return-to-play output combines acoustic recovery trajectory with comparator biomarkers to produce a multi-modal clearance indicator that can stand alongside blood panels and cognitive assessments in the clinical record.
Return-to-PlayMulti-Modal ClearanceClinical Record
05Technology

How Sona-2
Reads Voice

Five stages. Raw audio in. Clinically actionable triage output.

STEP 01
Audio Capture
Any recording device. Smartphone, tablet, sideline mic. Sona-2 is hardware-agnostic. Clean voice sample required — standardized prompt or free speech of sufficient length.
STEP 02
Feature Extraction
Audio decomposed into 88 acoustic features — frequency, energy, spectral, and temporal parameters — using the extended Geneva Minimalistic Acoustic Parameter Set.
STEP 03
Encoder
Features passed through a contrastive language-audio pretraining encoder backed by a hierarchical token-semantic audio transformer. Maps features into the LAM's learned embedding space.
STEP 04
Sona-2 LAM Inference
The Large Acoustic Model runs inference against its trained TBI phenotype. Acoustic signature matched against the model's learned representation of post-concussion voice patterns.
STEP 05
3-Tier Triage Output
Model returns one of three outputs: Elevated Risk, Monitor, or Clear — each with a confidence score. Designed for clinical integration, not consumer interpretation.
06Commercial

Built to License
Globally

Not a one-time sale. A recurring license to every professional sports organization that has athletes at risk of concussion — which is all of them.

The Licensing Approach
Global · Per-organization · Recurring
ModelAnnual per-organization license
DeploymentCloud-based, hardware-agnostic
IntegrationEHR and team medical staff workflows
DataDe-identified, federated — no raw audio leaves the org
ExclusivityNon-exclusive by default
ExpansionPer-sport, per-league volume pricing available
Amplifier's strategy is global licensing at scale — not acquisition or exclusivity deals.
Why Licensing, Not Selling
A one-time sale captures a fraction of the addressable value. Every professional contact sport franchise globally — rugby, football, soccer, hockey, MMA — faces the same concussion liability. A per-organization annual license scales with that market without giving up the platform.
The Market
Thousands of professional and elite sport organizations worldwide carry medical liability for athlete concussion. Each is a license target. The commercial motion begins with existing engaged partners and expands by sport and league.
What Unlocks It
A clean prospective dataset with blood biomarker and SCAT6 comparators. That's what the current study produces. Validation data is the commercial key.
07What's Next

Where This Goes
From Here

The study is underway. The commercial framework is in place. Three near-term milestones define the path to deployment.

01
Full Dataset Assembled
The prospective study runs 18 weeks with multi-modal biomarker collection across the acute and sub-acute recovery window. End of study delivers the first longitudinal voice biomarker dataset in confirmed concussion — with SCAT6 and blood panel comparators built in.
02
Sona-2 TBI Model Locked
Once the dataset is assembled, the Amplifier AI team runs validation analysis — sensitivity, specificity, and confidence calibration against comparator modalities. The TBI phenotype model is refined and locked for commercial deployment.
03
Global Licensing Opens
Validated model. Clean clinical evidence package. The licensing program opens to professional sports organizations globally — starting with existing engaged partners and expanding by sport and league.
Partner With Amplifier
The Fifth Vital Sign
Is Voice

Sona-2 is the first Large Acoustic Model built for clinical deployment. If your organization works with athletes at risk of concussion, we want to talk.

Request Information