Attribute Captured Knowledge to the Person Who Shared It.
Voice Profile is an optional, opt-in way to attribute captured knowledge to the person who shared it — with explicit consent, encrypted voice embeddings, and a clear audit trail. It is not biometric authentication and is never used to sign in.
A transcript tells you what was said. Not who said it.
A transcript tells you what was said. It doesn't tell you whether the person was who they claimed to be, whether they consented to being recorded as the source, or whether their contribution was attributed before their knowledge became organizational policy.
No speaker attribution
Without voice attribution, knowledge records have no clear connection to the person who provided them. Authorship is assumed, not recorded.
No consent trail
Recording knowledge without explicit consent is a governance gap. Voice Profile requires 3-checkbox explicit consent before any voice data is processed.
No attribution at capture time
Attribution established after the fact is weaker than attribution captured in the moment. Voice Profile operates at capture time — not retroactively.
Consent first. Encrypted always. Deletable on demand.
Voice Profile is designed with minimal data collection, maximal user control, and a clear audit trail at every step from enrollment to matching.
Admin-gated enablement
Workspace admin must enable Voice Profile. It is off by default at the workspace level and cannot be activated by individual users.
3-checkbox explicit consent
Each user sees three explicit consent checkboxes: voice data processing, encrypted storage, and right to delete. All three must be checked before enrollment begins.
Enrollment via 3 spoken phrases
Audio is processed server-side only. A mathematical embedding is derived and stored AES-256-GCM encrypted. Raw audio is never retained after processing.
Cosine-similarity matching
Matching returns confirmed, uncertain, or failed — never exposes a raw embedding to the browser. The similarity score is logged for audit purposes.
Attribution source on every article
Every captured knowledge article records the attribution source: account login, voice-matched, manual attribution, or unknown. The source is locked once approved and cannot be edited through the application UI.
User-controlled deletion
Users can disable or permanently delete their voice profile at any time. Deletion permanently overwrites the encrypted embedding — it cannot be recovered.
Chain of custody that starts at the moment of capture.
Chain of custody
Every approved article shows who spoke it, how they were attributed, and what confidence applied. The provenance trail starts at the moment of capture.
Audit trail
Consent events, matching events, and workspace settings changes are all logged. The audit log is append-only and cannot be edited through the application UI.
Governance evidence
Voice-attributed captures are flagged in the Compliance Evidence Map, contributing to the attribution domain in your audit-readiness coverage.
Questions about Voice Profile.
Is Voice Profile required?
No. It is off by default at the workspace level. Admins must enable it, and each user must explicitly consent before enrollment.
What voice data is stored?
A mathematical embedding (vector) derived from vocal characteristics. Raw audio is never stored after processing. The embedding is AES-256-GCM encrypted and cannot be decrypted by the browser.
Can Voice Profile be used to log in?
No. Voice Profile is not biometric authentication. It is never used as a factor in login or account access — it only affects voice-based attribution in knowledge records.
What happens if I delete my voice profile?
Your encrypted embedding is permanently overwritten before deletion. The attribution on historical articles is preserved as metadata but your profile cannot be restored.
Add voice attribution to your knowledge governance.
Voice Profile is available on the Operations plan. Enable it in workspace settings, invite your team, and start attributing expert knowledge to the people who shared it.