Privacy

Everything the app remembers lives in this browser, on this device. Nothing is synchronised to another device, and using SpeakClear elsewhere starts from scratch.

• Your audio — in IndexedDB, under the database speakclear.
• Everything else — in local storage, under keys beginning speakclear.: your first name if you entered one, your onboarding answers, completed lessons, quiz results, lesson notes, bookmarks, self-ratings, practice log, delivery scores, and display settings.

You can download all of it as JSON from your progress page, delete recordings individually or all at once there, or remove everything below.

The microphone is opened only when you press record, and it is released as soon as you stop. Nothing listens in the background. The delivery feedback — pace, pausing, volume, consonant crispness — is computed in your browser with the Web Audio API; the audio is never sent anywhere for analysis.

We ask your browser for the unprocessed microphone signal so the measurements reflect your voice rather than your browser's noise filtering. We do not perform voice identification, voice cloning, or any biometric matching, and we do not use your recordings to train anything.

The spoken lesson lines are synthesised by ElevenLabs through this site's own /api/tts endpoint and cached. Your browser requests a fixed line from the course script by its identifier — none of your speech, text or notes is involved, and the request carries nothing about you beyond what any web request carries (your IP address and browser user-agent, which the hosting provider logs as standard). If the studio voice is unavailable, playback falls back to your device's built-in voice, which never leaves the device at all.

To learn whether the course actually helps anyone, this build can send anonymous usage events to a third-party analytics service (PostHog or Plausible, depending on how it is deployed). This is off until you allow it, and you can change your mind at any time.

What an event contains, at most:

• which thing happened — a lesson opened or completed, a baseline recorded, a workout finished, the paywall seen
• plain numbers about it: a duration in milliseconds, a words-per-minute figure, a delivery score, which lesson slug, how many days since your first visit
• the page path and referrer, and a random ID stored on this device

What is never included: your recordings or any audio, your lesson notes, your transcripts, your self-ratings, your name, or your email address.

SpeakClear has no accounts and asks for no personal details to use it. The only place we invite an email address is the optional "would you pay for this?" form, and it is only sent if you type one in and submit it. It goes to the form endpoint configured for this deployment, along with anything you wrote in the note field and the price option you picked. We use it to tell you if a paid version becomes real. We do not sell it, and we do not add it to any list you did not ask for.

• Upload, store or listen to your recordings on a server.
• Use voice biometrics, voice cloning or speaker identification.
• Sell or share your data with advertisers or data brokers.
• Rank you against other learners or share your scores with anyone.
• Treat an accent as an error — we train intelligibility, not accent removal.

SpeakClear is currently free, with no payment processing of any kind, so no card or billing details are collected or stored.