Local file processing
When you choose a video, browser file APIs read only the portions required by the supported fingerprint formats. The OpenSubtitles calculation reads the first and last 64 KiB. The NapiProjekt calculation reads the first 10 MiB. The calculations happen in your browser, and those video bytes are not uploaded to SubtitleFinder.
Information used for subtitle search
The search API receives the calculated fingerprints, the video file size, filename, and your selected language. It forwards the lookup information required by OpenSubtitles and NapiProjekt. SubtitleFinder does not intentionally persist that search data in a user profile.
Accounts, sign-in, and billing
When you create an account, SubtitleFinder stores your email address, optional display name, a one-way salted password hash when you use email sign-in, and revocable session records. Google sign-in stores the Google account identifier and profile details supplied by Google. Passwords and Google access tokens are never stored in plain text.
Paid plans are processed by Stripe. SubtitleFinder stores Stripe customer and subscription identifiers, plan, billing interval, status, and renewal date so it can authorize features. Stripe processes payment details under its own privacy policy; SubtitleFinder does not receive or store full card numbers.
Product analytics and operational logs
SubtitleFinder records funnel events and completed-search logs to understand whether matching succeeds. These records can contain the selected filename, along with coarse properties such as video extension, a file-size range, language, result count, provider, and failure category. Funnel events do not contain file hashes, subtitle contents, or persistent user identifiers. Browser Do Not Track prevents the separate browser funnel events, but the filename remains part of the subtitle search request and completed-search operational log.
Optional contact and feedback messages
If you use the contact and feedback form, SubtitleFinder sends the message type, subject, message, and any name or reply email you choose to provide to the site owner. This information is used to review the report and, when you provide an email address, reply to you. No selected video, filename, fingerprint, subtitle, or audio is attached to the message automatically.
Optional subtitle translation
When no exact subtitle exists in your selected language, SubtitleFinder may offer translation of an exact English subtitle. The API downloads that exact subtitle and sends its cue text to Cloudflare Workers AI only after you select the action. For an English filename fallback, synchronization remains local; if you choose “Synchronize, translate and download,” the browser sends the resulting synchronized SRT to SubtitleFinder and its cue text is translated by OpenAI GPT-5 mini. The selected video and video bytes are never uploaded, and subtitle timestamps are not sent to either translation model.
Completed translated SRT files are cached in Cloudflare KV for up to 30 days so the same source subtitle and target language do not need to be processed repeatedly. The cache identity is derived from the media fingerprint, subtitle provider and file identity, target language, and processor version; the cached SRT is not added to your account profile. OpenAI states that API data is not used to train its models unless the customer opts in, while default abuse-monitoring logs may retain prompts and responses for up to 30 days.
Optional audio transcription
When neither the selected language nor English has an exact fingerprint match, SubtitleFinder may offer to create subtitles from the video audio. This starts only after explicit consent. FFmpeg runs in the browser, reads the selected video through a read-only file mount, and extracts one-minute mono audio chunks. The complete video is not uploaded.
Audio chunks are sent one at a time for Cloudflare Workers AI speech recognition. A chunk is staged temporarily in Cloudflare KV so durable processing can retry it, then deleted after successful processing; abandoned inputs expire within two hours. Detected speech text may be translated to the selected language, and each completed text response is cached in Cloudflare KV for up to 24 hours to make retries resumable. Cloudflare states that Workers AI customer content is not used to train AI models or improve its services without explicit consent.
Downloads and third-party providers
Subtitle searches and downloads depend on external subtitle providers. Requests to those services are made by the SubtitleFinder API. Their availability, coverage, and handling practices are governed by their own services and policies.