The platform

Deploying speech biomarkers in real-world studies

A configurable solution for the design, administration and remote conduct of studies collecting speech in medicine and pharmaceutical research, built for unsupervised use at home, as well as assisted and in-clinic collection.

Book a walkthrough →

Participant-facing

Meeting participants where they are

The participant-facing side prioritises comfort and ease, on a device the participant already has.

A man in an armchair reading a short passage aloud from his phone

No smartphone needed

Automated phone call

For participants less familiar with smartphones, or when only acoustic stimuli are needed. No setup, no installation, no smartphone.

With guidance

Assisted phone call

For participants who need human guidance through the assessment: the same protocol, administered by site staff on the line.

Visual stimuli

Browser and app

For participants who own and are comfortable with a smart device, and for tasks that need a visual stimulus: picture description, reading passages.

Cephalgo app screen: Sustain a vowel — **Sustain a vowel**One instruction per screen, with a progress bar across the session.

Cephalgo app screen: Repeat syllables — **Repeat syllables**A timed motor-speech task, recorded at the tap of one button.

Cephalgo app screen: Read a passage — **Read a passage**A standard passage, read aloud at a normal pace.

Cephalgo app screen: Noise check and retake — **Noise check and retake**Background noise is flagged during recording, with a retake before the task is finished.

For study teams

A command centre for study design and administration

A web interface for researchers and clinical teams to design a protocol, schedule and trigger assessments, and monitor collection as it happens.

Integration

Measures built in

Cephalgo's speech measures run inside the platform, so collection and analysis come in one procurement.

Scheduling

Real-time and scheduled collection

Trigger an in-clinic assessment on the spot, or schedule remote sampling between site visits at whatever cadence the protocol calls for.

Scale

Small studies to multi-centre trials

The same protocol definition runs across sites and languages, which is what keeps a multi-country study comparable.

CStudies / Parkinson's study / Participants

ParticipantsReadings on the MDS-UPDRS speech item

Participants4

Home sessions33

Accepted at capture32

Retakes due1

Participant AHome readingClinic scoreRetake

Participant	Sessions	Latest reading	Status

AnalysisParticipant A, home session 8

Reading on the speech item1.5

Score at last visit1

StreamsAgree

Features that drove this readingRanked by contribution

Pitch variability

Pause time

Consonant precision

Vowel articulation

Speech rate

Stream	Reading	Tasks used
Deep-learning embeddings	1.5	All four tasks
Acoustic-linguistic	1.4	All four tasks
Cognitive-linguistic	1.6	Spontaneous speech

ScheduleSet for each cohort

Session	Tasks	Length
Home, every two weeks	Sustained vowel · rapid syllables · reading aloud · spontaneous speech	A few minutes
Clinic visit	Same tasks, with the MDS-UPDRS speech item rating	A couple of minutes

QualityParticipant B, latest session

Check at capture	Result	Action
Background noise	Above threshold	Retake requested in the app
Clipping	Pass
Speech duration	Pass
Input route	Built-in microphone	Logged

A failed check asks for a retake while the participant is still in the session.

ExportsReadings, attribution, quality flags, versions

Delivery	Content	Status
CSV / Parquet	Every session, with device and analysis version	Ready
API endpoint	Readings as they are returned	Live
eCOA integration	Into the study’s existing system	Configured

Uncompressed audio is kept with each accepted recording.

Data handling

Secure handling of everything collected

Speech from a clinical population is special-category data under GDPR. The platform is built on that basis from the start.

Processing

EU processing and storage

Data is processed and stored in the EU. DPIA, ROPA and DPA are provided as part of onboarding.

Consent

Granular and withdrawable

Participation and recording are consented separately, and withdrawal stops collection; existing recordings are handled as the protocol and consent specify.

Quality

Documented development

Developed under a documented quality management system, for use in regulated research environments.

Contact

Tell us how your study runs.

The schedule of assessments, the sites, the languages, and whether participants have smartphones. We will tell you which front-end fits and how we run it.

Book a walkthrough →