Your recordings start with your permission

Where voice takes are stored

Accepted takes are saved in IndexedDB on the device and browser you use. When you are signed in and Supabase cloud sync is available, accepting a take also uploads that recording to private storage associated with your account. Signed-out takes remain on the device. The in-app save notice tells you where a take was saved.

Deletion and exports

Restarting a completed scene removes its local accepted takes for that scene. Clearing site data in your browser also removes local recordings. You can delete a saved transcription from the workspace to remove its database record and private TXT, SRT, and WebVTT files. A video you export is created for you to download and is then controlled by the storage and sharing choices on your device. Deletion of remotely synced voice takes is handled through the support process.

Usage limits

To enforce transcription fair use and other tools' sign-in and usage limits, SceneDub creates one random device identifier in local browser storage and a same-site cookie, then derives one-way server hashes from that identifier and limited network/browser signals. Supabase stores the hashes, account ID when signed in, tool name, UTC date, count, processing duration, estimated provider cost, and event status. It does not store the original IP address, browser identifier, uploaded media, search text, or transcript in the usage tables.

Transcription files

When you use the audio transcriber, the selected file is uploaded directly to a private, short-lived Supabase Storage location and Deepgram receives a temporary signed link for speech recognition. SceneDub deletes the source media after processing; abandoned upload tickets expire after 15 minutes and are cleared during subsequent upload maintenance. If you are signed in, SceneDub saves the generated transcript, timing data, basic source-file metadata, and private TXT, SRT, and WebVTT export files to your account. Signed-out results are not retained after the browser tab closes. Do not submit material you do not have permission to process.

Account details and saved work

Google verifies your identity through Supabase Auth. SceneDub stores your account ID, email address, display name, avatar URL, and a refreshable sign-in session so the saved workspace can identify its owner and keep the account signed in across visits. Database row-level security and private storage rules restrict profile records, transcript text, timing data, and generated files to the owning account. SceneDub does not request Google Drive access and does not receive your Google password.

Feedback and review prompts

After a signed-in user completes a tool action, SceneDub may occasionally ask for a rating and optional written feedback. Supabase stores the account ID, prompt type, rating, optional message, number of prompts, prompt and submission times, and a 30-day snooze time after a dismissal. Submitting feedback permanently marks both prompt types complete for that account, so neither is shown again. Feedback text and rating are available only in the private operations dashboard and are not sent to PostHog.

YouTube API Services

The meme finder uses YouTube API Services to retrieve public video metadata and displays videos through YouTube's embedded player. SceneDub does not request access to your YouTube account. Your use is also subject to the YouTube Terms of Service and Google Privacy Policy. YouTube and Google may receive technical information from your browser when their player or services load, according to their policies.

Site data and third parties

SceneDub does not include advertising trackers. The browser requests the scene media needed for playback and, when configured, communicates with the deployment's Supabase project for authentication and private recording storage. Deepgram processes transcription requests, while YouTube and Google provide search metadata and embedded playback. Links to original media sources and licenses open third-party websites governed by their own policies.

Analytics

SceneDub uses Vercel Web Analytics to measure aggregate visits and page navigation without third-party advertising cookies. SceneDub also uses PostHog for product analytics, limited browser error diagnostics, and, when enabled, privacy-masked session replay. PostHog may store an anonymous browser identifier so visits within a session can be connected. After sign-in, SceneDub uses only the internal account ID to connect product activity to that account; it does not attach the account name or email address. SceneDub masks form inputs, visible page text, and element attributes in replay, removes query strings and fragments from analytics event URLs and recorded network request names, respects browser Do Not Track settings, and does not intentionally send search terms, YouTube titles or URLs, uploaded media, voice recordings, or transcript contents through the analytics integration. See the Vercel Web Analytics privacy documentation and PostHog privacy policy for provider details.

Interaction measurements include click heatmaps, scrolling, repeated or unresponsive clicks, and browser performance metrics. Structured logs and counters identify which tool completed, not its input or output. Error diagnostics retain error types and source locations while redacting error messages. Replay excludes microphone and media elements, embedded players, canvases, sign-in screens, and the saved workspace; admin and authentication routes are excluded from analytics events. Network request bodies and headers, console output, clipboard contents, and campaign query parameters are not collected by this integration. The PostHog SDK can support feature flags, experiments, surveys, and product tours when configured, but its automatic survey popups are disabled so they do not duplicate SceneDub's existing feedback prompt.

Questions

Use the support page to report a privacy or data issue. Do not include private recordings, credentials, or sensitive personal information in a public issue.