Cadence

Behavioral signal from your own meetings — voice, gaze, facial dynamics, speech timing — captured locally on your Mac and grounded in 47 peer-reviewed studies.

Download for Mac Browser extension

Universal binary — Apple Silicon and Intel · macOS 26+. Extension works with Chrome, Brave, Arc, Edge, Opera.

What Cadence sees

A few hundred numbers per 30 seconds of conversation. None tell a story on their own; together they're a behavioral fingerprint.

Voice prosody

Pitch contour, jitter, shimmer, speaking rate, pause distribution.

Gaze patterns

Fixation duration, saccade frequency, time looking toward the camera.

Facial dynamics

FACS-12 action-unit intensities, expression entropy, head-movement RMS.

Speech timing

Turn-taking, latency to respond, balance of who is talking when.

Lexical content

Vocabulary diversity, repetition, sentence structure — never the words themselves.

Cross-session signal

The same 31 indicators extracted week after week, so trends are comparable.

Where it captures

Native meeting apps, browser meetings, and LMS-embedded sessions inside iframes.

Native macOS apps no extension needed

Zoom Microsoft Teams Webex FaceTime

Browser meetings Cadence Bridge

Zoom Web Google Meet Microsoft Teams Web Webex Web

LMS-embedded (iframe) Cadence Bridge

Canvas Conferences Blackboard Collaborate BigBlueButton Brightspace / D2L Moodle

What the Bridge does

Cadence Bridge is a small Chrome extension that solves one problem: meetings hosted inside browser iframes are invisible to macOS's window-level detection. The native app sees the parent Canvas URL — not the BigBlueButton call running inside it.

The Bridge hooks getUserMedia and RTCPeerConnection at the page world via a MV3 main-world content script, then forwards lifecycle events (candidate / connected / closed) to the Cadence app over a local Native Messaging channel. It never captures media. It never sends anything to a server. The Cadence desktop app does the actual capture, locally, with the same consent flow as native meetings.

How it reasons

Not one model. Eight specialist agents that cross-check each other.

Cadence fans the feature stream out to a panel of agents, each focused on a single signal. Their outputs are reconciled by a final Synthesizer pass, with explicit calibration anchors and citation traces attached to every indicator that surfaces in the dashboard.

AcousticAgent

F0, jitter, shimmer, pause structure.

GazeAgent

Fixation, saccade, camera-attention.

LinguisticAgent

Lexical diversity, disfluency, MLU.

TemporalAgent

Turn-taking, response latency.

BaselineAgent

Compares to your own calibration.

CitationAgent

Pulls supporting studies from the library.

ClinicalFramingAgent

Keeps language correlational, never diagnostic.

Synthesizer

Reconciles the panel into the indicator set the dashboard renders, with citations + repeatability chips attached.

Grounded in literature

Every metric carries a citation. Every citation can be resolved to an open-access abstract.

peer-reviewed studies seed the on-device citation library on first launch

indicators in docs/METRICS.md, each tied to its strongest support

∞

live retrieval via OpenAlex — DOIs resolve to abstracts the CitationAgent quotes from

When the CitationAgent surfaces an indicator, it doesn't paraphrase — it pulls the actual abstract from the local library (seeded with 47 studies, extensible via OpenAlex), quotes the relevant claim, and links the DOI in the dashboard. Admins can upload their own PDFs to the library for closed-access work.

Repeatability is gated separately. Indicators whose worst-case driver falls below the Fleiss 0.75 clinical-decision threshold (per Stegmann 2020 ICC + WSCV priors) get a "low repeatability" chip in the dashboard — they're not suppressed, but they're marked so reviewers see them in context. The repeatability harness ships with the app for local recompute.

Fusaroli 2017 · ASD voice meta-analysis (30+ studies) Bone 2014 · jitter / shimmer / HNR Parish-Morris 2016 · pause structure Losh & Capps 2003 · lexical diversity Stegmann 2020 · repeatability priors

Where it lives

All of this happens on your Mac. Raw audio, video, and transcripts never leave the device.

Cadence captures and processes meetings locally, with macOS permission prompts you can revoke at any time. The only thing that ever ships to a server — and only if you opt in — is a small numeric fingerprint of the features above. No recordings. No transcripts. No third-party SDKs or analytics.

Cadence is not a diagnostic device. Indicators are correlational with published research findings, not predictions about any individual. Built for opt-in deployment with diagnosed families, consenting adults, and supervised clinical / research use.