Introduction

Analyze voice recordings with the wellness model (`pulse`); other models are available.

The default path is submit audio → analyze → get results, using the wellness model (pulse). Other models are available. Walk through it once in the Quickstart.

Submit, analyze, get results

Submit. On the analyze request, send the file as multipart audio. Formats, size, and quality codes: Audio Requirements. If the bytes should not ride on that request, mint a PUT URL first — Audio Uploads.

Analyze. POST /v2/models/pulse/analyze returns a job immediately (status: queued, result: null) — Model API. Other bundles: Choosing a Model or Sign.

Get results. Poll GET /v2/jobs/{job_id} or use a webhook until status is done. On a model job, route on result.summary.recommended_action — Interpreting Results.

As-is

All models and API outputs are provided as-is. Labels such as Established, Emerging, and Investigational describe published research depth, not a commercial tier. See Sign Catalog — evidence tiers. Outputs are biomarker signals, not diagnoses.

Advanced

• Specific sign analysis — Result for one health sign from the recording.
• Longitudinal analysis — Analysis that takes the person’s past results into account.
• Group-based analysis — Group-level or population-level aggregate analysis.

Where to go next

• Jobs, Polling, and Webhooks — job statuses, polling cadence, and webhook signatures.
• API Reference — base URL, auth, rate limits, endpoint map.
• Production Checklist — ship against this API.
• Support — FAQ and how to reach the team.