Say It Right — Privacy Policy
Last updated: 2026-10-03
This privacy policy applies to the mobile app "Say It Right" on Android and iOS:
- Google Play: package com.moon.pronounce, developer "janey-moon"
- Apple App Store: bundle ID com.moon.pronounce, seller "Heekyung Moon"
The app is developed and operated by Heekyung Moon (Google Play developer name: janey-moon).
Contact: ansgml2100@gmail.com
OVERVIEW
Say It Right does not require an account, and we do not sell personal information. You type a word or phrase, hear it in your phone's text-to-speech voice, say it, and see what the phone's speech recognizer heard, together with your recording and a pitch shape comparison. Your phrases, practice history, and recordings are kept in the app's storage on your device. The app also uses Google services for analytics, crash reporting, remote configuration, and advertising, and, only after you allow it, an online speech recognition service.
MICROPHONE, SPEECH RECOGNITION, AND TEXT-TO-SPEECH
The app asks for microphone access (and on iOS, speech recognition access) when you first tap Say it or Record, not at launch. If you decline, the app shows a message, and you can still type phrases and listen to them.
Recognition on the device: when the phone can enforce on-device recognition for the language (iOS with on-device support, or Android 13 and later), the app tries that first and requires on-device processing for those attempts.
Online recognition: for other languages or phones, recognition needs an online service: Apple on iOS, or on Android the speech recognition service your phone uses (for example Google or your phone maker). Before the first online attempt, the app names the service and asks you. If you choose Allow, the audio of that attempt, and of later attempts in languages that need online recognition, is sent to that service, which returns the recognized text and handles the audio under its own terms. If you choose Not now, recognition stays off for those languages, and you can still listen and record. Your choice is saved per service and can be changed or revoked at any time in Settings & privacy.
Recordings: when the phone supports it (iOS, Android 13 and later, or the Record mode on any phone), each attempt is saved as an audio file in a recordings folder inside the app's storage, so you can play it back. Listening stops after about 20 seconds; Record mode keeps at most 15 seconds. While recognition runs, the phone may first write the audio to the app's cache before the app moves it into that folder. If an attempt finishes with nothing recognized, its recording is deleted. If an attempt is cancelled (for example when you leave the app), a recording of it may remain in the app's cache folder; the system can clear that folder, and clearing the app's data or deleting the app removes it. The app does not upload recordings itself, and recordings are not included in analytics or in exported backups.
The reference voice comes from your phone's text-to-speech engine. To draw the pitch comparison, the app renders the phrase with that engine into a temporary audio file in the app's temporary or cache folder (iOS replaces it with each new render; Android keeps the 10 previous renders plus the current one) and compares it with your recording within the app. Depending on the voice and engine on your phone, the text-to-speech provider may process the phrase text under its own terms.
INFORMATION STORED ON YOUR DEVICE
My Phrases: up to 2,000 phrases, each with its text, language, optional list, tags, and created/updated times; up to 100 lists, each with a name.
History: up to 500 attempts (newest first). Each holds the phrase text, language, time, the text the recognizer heard (or none for Record mode), the word or character counts that matched or were missed, whether recognition ran on the device, and a link to its recording file. When an attempt falls off the end of the history, its recording is deleted.
Settings: practice language, voice speed, chosen voice per language, your online-recognition choice per service, the dates on which on-device recognition was confirmed or found unavailable for a language, and whether the introduction was shown. If saved data cannot be read, a copy is kept on the device so it is not silently lost.
The app also stores basic operating information: app-use count, whether the review prompt was shown, recent full-screen ad times, and cached advertising and More Apps configuration.
The developer does not run a server that receives your phrases, history, or recordings. The app reads developer-controlled advertising settings and a More Apps list (with app icons loaded from the web) from Google Cloud Firestore; it does not write your data to Firestore.
ANALYTICS, CRASH REPORTING, AND ADS
Google Firebase Analytics may process limited app-usage and technical information, such as app launches, sessions, screen and menu use, app version, device type, and app-instance identifiers. Practice events record only simple values: that you played the reference voice or saved a phrase, whether recognition was off, on-device, or online, whether the transcript matched (yes/no), that on-device recognition was confirmed, your online-recognition choice, the result of a model download request, backup export or restore, and that unreadable saved data was recovered. Other events record the app-use count, your tracking-prompt answer, review prompts, and opening the menu, a screen, the privacy policy, or More Apps. They do not include your phrases, transcripts, tags, language, or audio.
Google Firebase Crashlytics may process crash reports and related technical information needed to diagnose app problems.
The app shows ads through Google AdMob: app-open ads at launch and on return to the app, interstitial ads after a matched result, banner ads on the My Phrases and History screens, and a native ad in the exit dialog. The app does not start a full-screen ad while it is listening or recording, or while the online-recognition question is open; full-screen ads can appear at other times. Google may process advertising identifiers and device, app, consent, network, and ad-performance information to provide and measure ads, and may use an IP address to estimate an approximate location. The app does not request location permission. Ad revenue events (ad format, value, currency, precision) are recorded in Analytics.
PERMISSIONS AND ADVERTISING CHOICES
Feature permissions: microphone, and on iOS speech recognition. The app does not request camera, location, contacts, photos, or notifications.
Where required (for example in the EEA and UK), Google's consent message (UMP) is shown first, and personalized advertising stays off unless your choice allows it. On iOS, the App Tracking Transparency prompt is shown only after that step and only when personalized advertising is allowed. If tracking is not allowed, the IDFA is not available to the app or its ad partners. If Google's consent step fails at launch (for example, the consent message cannot be loaded), the app may still request non-personalized ads and does not show the tracking prompt at launch; the prompt can appear later only if you open Privacy choices and allow personalized ads. The app works whatever you choose.
You can change microphone, speech recognition, and tracking permissions in your device Settings at any time. Where Google requires it, the app menu shows "Privacy choices", which reopens the consent message so you can change your advertising choice.
STORAGE AND YOUR CONTROL
On My Phrases, you can delete a phrase from its edit sheet, or delete a list (its phrases are kept). On History, Clear history deletes all attempts and their recordings. In Settings & privacy, Delete all recordings removes the recordings saved with your history attempts but keeps transcripts and phrases (a recording left in the cache folder by a cancelled attempt is not covered), and you can revoke online recognition. Restore from backup replaces My Phrases and History with a backup you paste in. There is no in-app delete for a single attempt. You can also clear the app's data on Android or delete the app.
Export backup writes a file with your phrases, lists, and history (no recordings) to the app's cache folder and opens the share sheet; that file may stay in the cache, and copies you share are kept wherever you send them. Local data, including recordings, may be included in an Apple, Google, or device backup depending on your backup settings (Android app backup is not turned off for this app); uninstalling normally removes the active copy but may not remove an older backup.
Because we do not receive your phrases, history, or recordings, we cannot delete them for you; the controls above remove them from the app. For questions about data processed by Google or by a recognition service, or for help with a privacy request, contact us at the address below.
We use these services under their applicable terms, and we ensure that third-party service providers with whom we share user data provide the same or equal protection of that data as stated in this policy.
- Google Privacy Policy: https://policies.google.com/privacy
- Firebase privacy information: https://firebase.google.com/support/privacy
- Google advertising technologies: https://policies.google.com/technologies/ads
- Apple Privacy Policy (speech recognition on iOS): https://www.apple.com/legal/privacy/
SECURITY
Your saved data and recordings are kept in the app's private storage on your device, which the operating system protects from other apps; they are not separately encrypted by the app, so protect your device with a passcode. Apart from the limited analytics events described above and audio sent to a recognition service you allowed, the app does not upload your saved data. Communication with Google services uses encrypted connections (HTTPS/TLS). No method of storage or transmission is completely secure.
IMPORTANT NOTICE
"Recognized as" shows what the phone's speech recognizer wrote down, and recognizers can mishear or autocorrect. Matches, word counts, and the pitch shape comparison are automated estimates and visual aids. They are not a pronunciation grade, a language-proficiency assessment, or speech therapy, and a matching transcript does not mean native-like pronunciation.
CHILDREN
The app is not directed to children under 13, and we do not knowingly collect personal information from children.
CHANGES
We may update this policy if the app or its services change. The date above shows the latest revision.
CONTACT
Heekyung Moon (Google Play developer: janey-moon)
Email: ansgml2100@gmail.com