Voiceprint (Sensitive Information) Processing Notice
Reference translation. The Korean original prevails in case of any difference.
Effective date: October 11, 2026 · Version 1
A voiceprint in SAYNARY is a feature that remembers a person's voice characteristics once a speaker has been named in a recording, and automatically adds the name when the same voice appears in later recordings.
Because a voiceprint is information about physical and behavioral characteristics created by technical means to identify a specific individual, it is biometric information — that is, sensitive information — under Article 23 of the Personal Information Protection Act and Article 18(3) of its Enforcement Decree. The Company therefore creates and uses voiceprints only with the separate consent of the owner of the voice. Even without consent you can keep using recording, transcription, minutes and all other features.
1. Basic Principles
| Principle | Details |
|---|---|
| Off by default | In an organization, the feature can be used only after an organization admin has reviewed the legal notice and turned it on. In a personal workspace you can enroll only your own voice |
| Separate consent | Consent is obtained from the owner of the voice separately, not bundled with other consents (Terms, Privacy Policy) |
| Processed only on the Company's servers | Voiceprints are created and compared only with the speaker recognition model on the Company's servers (Republic of Korea). They are never sent to external AI (such as Claude) or to other companies |
| No audio stored | Only features (embeddings) that turn the voice into numbers are stored. The original audio cannot be reconstructed from a voiceprint |
| Encrypted storage | Features are stored encrypted with the server master key |
| No other purposes | Used only for automatically naming speakers and suggesting names within the same workspace. Never used for entry control, payment, identity verification, surveillance or emotion analysis, and cannot be exported or downloaded |
2. What Is Processed
| Item | Details |
|---|---|
| Purpose | Recognizing the speakers in a recording and adding or suggesting their names automatically |
| Items processed | Voice features (256-dimensional embeddings; up to 50 samples and up to 50 counter-examples marked "not this person"), name, account id if an organization member, consent method, time, consent text version and note, number and length of samples, number of recognitions and last recognition time, the person who enrolled it |
| How it is created | ① Choose "Remember this voice" when naming a speaker in a recording (at least 20 seconds of clean speech by that speaker) ② Settings > Voiceprint > "Enroll my voice" (read a passage for 15–60 seconds; the audio is deleted right after processing) |
| Continuous learning | Tapping "That's right" on an automatically added name adds a sample (replacing the oldest ones), and tapping "That's not right" stores a counter-example so the same mistake is not repeated |
| Retention period | Until deleted or consent is withdrawn. However, it is deleted automatically if unused for the period set by the organization after its last use (1 year by default, 30 days to 10 years). If the organization turns the feature off, recognition stops immediately and voiceprints are deleted after 30 days (restored if turned on again within that time). On account deletion, deleted after the 7-day grace period |
| Provision to third parties, outsourcing, overseas transfer | None |
3. Whose Consent Is Obtained, and How
| Owner of the voice | How consent is obtained |
|---|---|
| Yourself | When enrolling, you review and accept the consent text below in the app |
| A member of the same organization | When enrollment is requested, a consent request is sent to that member as an app notification. It is never used for recognition before the member accepts, and it is deleted immediately if they decline |
| An external person (customer, business partner, etc.) | The user who enrolls the voice must inform the owner of the voice of this notice directly and obtain their separate consent, then confirm "I obtained the person's consent" and leave a note on when and how it was obtained (kept in the audit records). Possible only if the organization allows this method |
Falsely confirming that an external person consented violates the Terms of Service and may give rise to legal liability. The user who enrolls the voice should keep the record of consent (messages, written consent, etc.).
4. Your Right Not to Consent and Its Effect
You have the right not to consent to voiceprints. If you do not consent, names simply will not be added automatically by voice; you can use recording, transcription, speaker separation, minutes, search and all other features in the same way and name speakers yourself.
5. Checking, Deleting and Withdrawing Consent
- Under Settings > Voiceprint you can see the status of your voiceprint, the number of samples, the number of recognitions and the last recognition time.
- You can withdraw consent and delete it immediately at any time with "Delete my voice data". The person who enrolled it and organization admins can also delete it. After deletion, automatic recognition stops (names already added to recordings remain and can be edited by hand).
- External persons may ask the user who enrolled them, or the Company (support@metavu.io), to delete their voiceprint.
- Enrollment, deletion and automatic recognition are kept in the audit records.
6. Consent Text
The app shows the following text and asks for consent (consent text version 1).
[Optional] Consent to the collection and use of a voiceprint (biometric information)
주식회사 메타뷰 creates and uses voiceprints as follows. This consent is separate from other consents, and you can use the Service's other features even if you do not consent.
- Purpose of collection and use: Recognizing my voice in recordings and automatically adding or suggesting the speaker name
- Items collected: Features of my voice (the original audio is not stored), name, consent record, recognition records
- Retention and use period: Until I delete it or withdraw consent. Deleted automatically if unused for {보관 기간} days after last use
- Where it is processed: Only on the Company's servers (Republic of Korea); never sent outside or provided to other companies
- Right to refuse consent: You may decline; declining has no disadvantage other than names not being added automatically by voice
- You can delete it (withdraw consent) at any time under Settings > Voiceprint
☐ I have reviewed the above and agree to the creation and use of my voiceprint.
Confirmation when enrolling an external person
☐ I informed the owner of the voice ({이름}) of the purpose, items, retention period and right to refuse described above and obtained separate consent. (When and how it was obtained: {메모})
7. Contact
Please send inquiries about voiceprints to the chief privacy officer (노진송 (대표이사), support@metavu.io). Everything else follows the Privacy Policy.
| Version | Effective date | Description |
|---|---|---|
| 1 | 2026-10-11 | First version |