Privacy policy
Last updated: October 28, 2026
The short version: captions, the phrase board, lip tracking and practice run on your device, and we don't record or save your video or audio. The only time video leaves your device is the optional Record & guess research demo, after you confirm consent. If you join the waitlist, we keep your email only to tell you about MouthRead. We never sell your information.
Who we are
MouthRead is operated by PDX Flower Guy LLC, an Oregon limited liability company in Portland, Oregon, USA. Contact: hello@mouthread.com.
What stays on your device
- Lip tracker (web demo): uses your camera only after you press start and your browser asks permission. Video is processed inside your browser and is not recorded, saved or sent to us. The camera switches off when you press Stop, leave the tool or close the page.
- Offline captions page (/captions/) with the Vosk or Whisper engines: your microphone audio is turned into text on your device. The speech model files are downloaded once (Vosk from mouthread.com; Whisper from Hugging Face and jsDelivr) and then cached by your browser. Audio is not sent anywhere.
- Phrase board and practice: run on your device. Text-to-speech uses your device's built-in voices.
- Lip phrases (beta) (/lips/): the camera picture is processed on your device by MediaPipe and is never recorded or uploaded. For each phrase you train, a short list of lip-shape measurements (numbers such as mouth opening and width over time, not images) is saved in your browser's storage on this device. You can export it to a file you keep, or delete it any time with "Delete all training".
- Your text-size and similar settings are saved in your browser's local storage on your device.
When audio goes to your browser's speech service
- The captions in the lip tracker, and the "Chrome speech" engine on the captions page (used only if you choose it, or if the on-device engines can't start or are too slow), use your browser's built-in speech recognition.
- That service is run by your browser's maker, not by us. Chrome may send the audio to Google, and Safari may send it to Apple, to turn speech into text. Their privacy policies apply.
- We don't receive or keep that audio or the captions. Captions live only on the page and disappear when you close it.
Record & guess (optional research demo)
- Records a short clip of video only, with no sound, and only after you tick a box confirming the person on camera knows and agrees.
- The clip is uploaded to our guess server, which runs on Hugging Face (a cloud service in the USA). Hugging Face's infrastructure handles the connection, so it can see technical data such as your internet address.
- The clip is used once to make guesses and is deleted right after, even if there's an error. Our server code does not log the video, your internet address or the guess text. We don't keep the clip, share it, or use it to train anything.
- This feature is a free, non-commercial research demo on the website only. It is not available in the MouthRead app.
The waitlist
- If you join, we keep your email address, the date you joined and, if you choose one, your answer to "what best describes you" (a general, non-health question such as "I want captions for myself").
- To block spam and repeated sign-ups, the waitlist also stores your browser type (user agent) and a scrambled, one-way hash of your internet (IP) address. We never store the IP address itself, and we don't use these to find your location or track you.
- We use it only to send occasional emails about MouthRead, like news, early versions to try, and requests for feedback.
- Sign-ups are stored privately on our web host (Hostinger). A notice of each new sign-up is emailed to our own inboxes (Hostinger mail and Google Gmail).
- Every email lets you unsubscribe, or email hello@mouthread.com and we'll delete your details.
What we never do
- We never sell, rent or share your personal information for advertising.
- We don't use advertising trackers or analytics tools on this site.
- We don't try to identify the people who appear on camera, and we don't build face or voice profiles.
Other services the site uses
- Hostinger hosts this website and our email. Like any web host, it keeps basic visit records, such as internet address and browser type, to keep the site running and safe.
- The lip tracker downloads Google's open-source MediaPipe files from jsDelivr and Google's file servers. The captions page may download Whisper files from Hugging Face and jsDelivr. Those services see that a download happened, but they don't get your video or audio.
Children
MouthRead is not directed to children under 13, and we don't knowingly collect personal information from them. If you think a child gave us their email, write to us and we'll delete it.
Your choices and rights
You can ask to see, correct or delete what we hold about you by emailing hello@mouthread.com. We'll answer within 30 days. Depending on where you live (for example Oregon or California), you may have extra privacy rights; we'll honor them.
Security
The site uses HTTPS, and waitlist data is kept in a folder that isn't reachable from the web. No system is perfectly secure, so we keep as little as we can.
Changes
If we change how we handle information, we'll update this page and the date at the top. If it's a big change, we'll email the waitlist.