Security and your data
A voiceprint is biometric data in UK and EU law, so this is what Nanoru does with yours.
Voiceprints
- They are yours
- Each customer’s voiceprints are kept apart. They are never pooled or shared across customers.
- Only from your recordings
- Nanoru keeps a voiceprint for each voice it hears in the recordings you send, named or unknown, so it can recognise that voice again. It never builds one from audio you did not send, and we keep no library of public figures.
- Deleted when you say
- An owner can delete a person’s voiceprint in the dashboard or over the API, and Nanoru stops recognising that voice straight away. Lines already credited to the person keep their name until you delete the recordings.
- You are the controller
- You decide whose voices are named, and you are responsible for telling those people and for any consent the law requires. Nanoru processes the data on your instructions, under a data processing agreement.
How long things are kept
These are the defaults, and a contract can shorten them.
| What | Kept for |
|---|---|
| The original files you send | 7 days after the episode is ready |
| The audio from them, unchanged, and a smaller copy for playback | 30 days after the result is delivered |
| Transcripts, lines and words | The length of your contract, then deleted within 30 days |
| Voice measurements for every line, used to name a speaker in recordings already processed | The length of your contract, then deleted within 30 days |
| Fingerprints of repeated audio, used to find adverts and jingles. They cannot be played back | The length of your contract, then deleted within 30 days |
| Voiceprints | Until you delete them, or 30 days after your contract ends |
| Records of who changed or exported what | Two years |
Where data goes
Stored and processed in the EU
- Audio and video
- Voiceprints
- Transcripts
- Speech processing
Sent to Anthropic’s Claude API, outside the EU
- Transcript text
- The recording’s title and description
- Your shows’ names, descriptions and recent titles
- Names of people already named
- The questions you put to Ask, and its answers so far
Nanoru sends that text for three purposes: to suggest names for unknown speakers, to work out which show a recording belongs to when it arrives without one, and to answer the questions you put to Ask. It sends no audio and no voiceprints, and Anthropic’s terms say it does not train its models on the text. An owner of your organisation can switch all three off, and then none of your transcript text is sent.
Access and records
- Encryption
- Data is encrypted in transit and at rest.
- Accounts
- Each person has their own login and role, and can sign in with a passkey or through your own single sign-on. Only owners and admins can delete voiceprints or create API keys.
- API keys
- You issue keys for your organisation, limit what each one can do, and revoke them when you need to.
- In the browser
- Nanoru's pages are sent only over HTTPS and can't be shown inside another website. This site runs no scripts from anyone else.
- An audit log
- Every edit, deletion and export is recorded. Nanoru staff can open your transcripts, speakers and audio to support you, and your log records what they read, play and change.
- Certifications
- Nanoru has no security certifications yet. We will answer a security questionnaire and share our list of sub-processors on request.
Test it on your own recordings
Send 20 episodes or 20 hours of recordings. We return them with your presenters named, at no charge.