Analyze Voice
The core endpoint. Submit a few seconds of patient speech and Voxtar extracts sub-clinical neurological and respiratory voice biomarkers, scores them against the patient's baseline, and returns structured diagnostic signals. Audio is processed and discarded — only derived, pseudonymized features are retained. Accepts a hosted audio_url or a direct multipart file upload.
Headers
Body Parameters
The patient this sample belongs to. Drives baseline scoring and longitudinal tracking.
The voice sample. Either an audio_url to a hosted file, or a multipart file field. WAV/FLAC/MP3/OGG, 16kHz+, 3–60 seconds. A sustained vowel or read passage works best.
Which biomarker panels to run. Defaults to all enabled for your workspace.
Optional elicitation hint: sustained_vowel | read_passage | spontaneous | cough. Improves feature extraction.
When true, emit a FHIR Observation bundle to your configured EMR endpoint on completion. See Integrations → FHIR.
Response
{"id": "an_5Tz7Lm","patient_id": "pt_8kQ2v9","status": "complete","duration_ms": 812,"panels": {"respiratory": {"score": 0.62,"baseline_z": 2.8,"drift": "elevated","features": { "shimmer": 0.071, "hnr_db": 11.2, "expiratory_ratio": 0.38 }},"neurological": {"score": 0.14,"baseline_z": 0.3,"drift": "nominal","features": { "jitter": 0.009, "pause_ratio": 0.12, "artic_rate": 4.9 }}},"flags": [{ "code": "respiratory_drift", "severity": "review", "panel": "respiratory" }],"created_at": "2026-07-11T09:26:41Z"}