Privacy Policy

Effective 2026-08-27.

Transcriptinator ("the app") is developed and published by Cherry Blossom Development, LLC ("we", "us"). This policy describes what data the app processes, where it's stored, and when — if ever — anything leaves your device.

Short version

  • Audio never leaves your device.
  • Transcripts never leave your device unless you configure a third-party AI provider with your own API key.
  • We do not collect, store, or transmit any of your data on our servers. We don't run servers for this app.
  • The app has no accounts, no sign-in, no analytics, no advertising, and no tracking.

What the app processes

Audio recordings

When you tap Record or import an audio file, the app stores that audio file in its private data directory on your device (iPhone, iPad, or Android). It runs speech recognition, voice activity detection, speaker segmentation, and speaker embedding entirely on your device using open-source machine learning models (Whisper, Silero VAD, pyannote segmentation, ERes2Net) via the sherpa-onnx runtime.

Audio files are never uploaded to us or to any third party by default. Deleting a recording in the app removes both the audio and its transcript from your device.

Transcripts

Transcripts produced by the on-device speech models are stored in the app's private database. Deleting a recording removes its transcript.

Speaker voice profiles

When you enroll yourself or another person in the app, the ML embedding of their voice — a fixed-length numerical vector — is stored locally with the display name and colour you assign. This vector cannot be reversed back into audio. It's used only to auto-label speakers in your future recordings.

Preferences

Your preferences (auto-transcribe on/off, auto-classify on/off, preferred ASR quality tier, onboarding state) are stored in the app's private settings storage on your device.

Optional: sharing transcripts with a third-party AI service

The app can optionally send the plain-text transcript of a recording to one third-party AI service you select on the AI Providers screen. This is off by default. Nothing is sent until you complete three steps:

  1. Configure a provider (Anthropic, OpenAI, or Google) on the AI Providers screen.
  2. Enter your own API key issued to you by that provider.
  3. Toggle the consent switch to on and accept the in-app disclosure that names the provider and describes what will be sent.

What is sent, and to whom

When the feature is used, the app makes an HTTPS request from your device directly to the provider's API endpoint, authenticated with your key. The request contains:

  • The plain-text transcript of the recording you asked to classify. Transcripts may include names, places, and any other personal details you spoke.
  • A system prompt asking the provider to return structured items (todos, calendar events, reminders, notes, decisions, questions).

The following are never sent to any third party: audio recordings, voice embeddings, speaker profiles, your API key (it authenticates the request but is not part of the payload), or any preference or configuration data.

Recipients and their privacy policies

Depending on which provider you select, the recipient of your transcript is one of the following legal entities. Each provider handles your data under their own privacy policy, which offers protections we consider substantively equivalent to those described in this document, including confidentiality, security controls, and user rights to deletion:

Review the policy for whichever provider you enable before you share transcripts with them. Their data-retention, training, and access practices are governed entirely by their agreements with you — not by us.

Our role

We — Cherry Blossom Development, LLC — do not host, proxy, log, retain, or store transcripts sent to any AI provider. The traffic goes from your device directly to the provider's API. We have no server-side infrastructure for this app.

Revocation

You can revoke consent any time by toggling the switch off on the AI Providers screen. Once off, no further transcripts are sent. Changing providers also automatically clears prior consent — you must re-consent for the new recipient before any transcripts are shared with them.

Network activity

The app connects to the internet in only these situations:

  • Downloading transcription models from their official public sources during first-run setup or when you install a new quality tier from the Models screen. These downloads are HTTPS requests to the sherpa-onnx and Whisper release URLs on GitHub and Hugging Face.
  • Sending transcripts to the AI provider you configured (only if you enabled that feature).

There are no analytics beacons, no crash-reporting services operated by us, no advertising SDKs, and no third-party trackers.

Permissions

  • Microphone. Required to record. iOS shows the orange microphone indicator whenever the microphone is active; Android shows a system-level microphone icon.
  • Notifications (optional). On Android, used only to keep the recording indicator visible while the app is in the background. Denying this permission has no effect on recording quality.
  • Calendars (write-only). Only requested if you tap "Add to Calendar" on a calendar-event extraction. Used to open the system event editor pre-filled with the event details. Denying this permission simply disables that button.
  • System alert window / draw over other apps (Android, optional). Only requested if you enable the floating record button in Settings. Used to show the button on top of other apps so you can record without switching to Transcriptinator. Denying disables that feature.

The app does not request Contacts, Photos, Location, or any other permission.

Children's privacy

The app is rated 4+. We do not knowingly collect personal information from anyone, including children under 13. Because the app has no accounts and does not transmit personal data to us, there is no data collection to opt out of.

Data retention

On-device data (audio, transcripts, voice embeddings, preferences) is retained until you delete the specific recording, delete the speaker profile, or delete the app. There is no automatic expiry — the app does not phone home to check whether data should be removed.

Data you send to a third-party AI provider is retained according to that provider's policy and your account with them. We have no control over their retention.

Your rights

Because your data lives on your device, you always control it. To remove all of it: delete the app. iOS and Android remove the app's private data directory when you uninstall. To remove data you previously shared with a third-party AI provider, contact that provider directly using the account associated with your API key.

Changes to this policy

If we update this policy, the effective date at the top will change. We recommend checking back before you enable optional AI features so you know what's in scope.

Contact

Questions or concerns about this policy? Email privacy@cherryblossomdevelopment.com .