Fb Video Downloader has been removed from Google Play. We’re sorry — read what happened and how we’ll make it right

Sonary: MP3 & Noise Remover AI

Privacy Policy

Privacy policy for the Sonary: MP3 & Noise Remover AI Android app by NewAgeDevs.

Effective date: 2 November 2026

Last updated: 2 November 2026

App: Sonary: MP3 & Noise Remover AI (Google Play package com.newagedevs.sonary; Apple App Store ID 6798929819)

Developer: NewAgeDevs (Md. Imam Hossain), Bangladesh

Contact: imamagun94@gmail.com

This policy explains what Sonary does with information: what stays on your phone, what leaves it, who receives it and why, and the choices you have. It replaces the version dated 4 October 2026.

The short version. Your audio and video files are processed on your device by every standard tool (extract, convert, trim, split, merge, mix, equalize, speed, volume, compress, record) and by Offline AI, and never leave it. Your files, the text you type and your voice leave your device only when you start a Cloud AI job (for example a cloud vocal separation, an AI Voiceover or Clone my voice), and only after the app has asked your permission for that kind of data — or when you choose to include a voiceover’s text in a report. They go over an encrypted connection to our backend and on to our AI provider, Replicate, for that one job. We keep no copy of your audio or your voice recordings on our servers, and no copy of your voiceover text unless you include it in a report, and Replicate deletes what is left of each job within about an hour. Clone my voice is for your own voice only: you read a consent sentence aloud first, and we keep a record of that consent, not the recording. The app creates an anonymous account automatically so it can keep track of your AI minutes; you never have to give us your name or email, though you can choose to sign in with Google or Apple to back your minutes up. The free version shows ads from AppLovin. We never sell your data.

1. Who we are

Sonary (“Sonary”, “we”, “us”) is a mobile audio toolkit published by NewAgeDevs. It lets you extract audio from video, convert between formats, trim, split, merge and mix audio, adjust equalizer, speed and volume, compress and record audio, and run AI features: vocal/instrument separation, background-noise removal and the app’s other AI audio tools, AI Voiceover (text-to-speech) and, for Sonary Pro subscribers, Clone my voice (speech in your own voice). This policy explains what data the app handles, why, who processes it on our behalf, and the choices you have.

This policy covers the Sonary app for Android and iOS and the backend service the app talks to. It does not cover third-party services you may open from the app (for example, if you share an exported file to another app), which have their own policies.

2. What we collect and why

We collect the minimum needed to make the app work, keep track of the AI minutes you have, keep the app stable, keep the AI voice features from being misused, process purchases, and (for the free tier) show ads. Here is everything, grouped by purpose.

2.1 Audio, video & media you choose

  • What: audio files, video files, and voice recordings that you select or record so the app can process them. (The voice sample you record for Clone my voice is covered in Section 2.8.)
  • Where it is processed: on your device for every standard tool (extract, convert, trim, split, merge, mixer, equalizer, speed, volume, compress, voice recorder) and for Offline AI. For these, your media never leaves your device. On-device processing uses FFmpeg and an on-device AI runtime (ONNX Runtime).
  • Cloud exception: if you choose Cloud AI for a tool that works on an audio file (such as vocal separation, noise removal, Studio Voice, Stem Splitter or Subtitles), the audio for that one job is uploaded for processing (see Section 3). If you picked a video, its audio track is extracted on your device and only that audio is uploaded — never the video.
  • Why: to perform the exact task you asked for.

2.2 Your Sonary account

  • Anonymous account (automatic). The first time the app starts, it signs in anonymously with Firebase Authentication (Google). This creates a random account ID with no name, email or other details attached. You don’t register or type anything.
  • Why: your AI-minute balance, any subscription allowance and your Cloud AI jobs are kept on our server under this ID, and so are your voice-consent records and any reports you send, if you use those features (Sections 2.8 and 2.9). Every request the app makes to our server carries a short-lived, signed token that proves the ID, so a balance can’t be forged or spent by anyone else.
  • App integrity check. Requests to our server also carry a Firebase App Check token, issued through Google Play Integrity on Android or Apple DeviceCheck on iOS, which helps our server tell the genuine Sonary app from scripts and modified copies. Our server keeps only the result (valid, invalid or missing) with your Cloud AI job records and voice-consent records.
  • Optional sign-in. In Settings you can back up your minutes by signing in with Google (Android and iOS) or with Apple (iOS). If you do, Firebase Authentication links that sign-in to your account ID and stores what the provider shares with it: your email address (or an Apple “Hide My Email” relay address), the provider’s identifier for your account and, where the provider supplies it, your name. The app shows the email in Settings so you can see which account holds your minutes. We don’t use it for anything else — no marketing email, no newsletters.
  • Why sign in: so your minutes survive reinstalling the app and can be restored on another device. The first sign-in with a given Google or Apple account also adds a one-time bonus of free minutes.
  • Undoing it: “Remove backup” in Settings unlinks the Google or Apple sign-in and keeps your minutes on this device; “Delete account” deletes the account and its server data (see Section 9).

2.3 Purchases & subscriptions

  • What: which subscription or minute pack you bought, the store’s purchase token for it, whether a subscription is active, when it renews or ends and whether it is set to renew, and the minutes it added to your account. Payments are handled by Google Play or the Apple App Store — we never see or store your card number or billing details.
  • How: the store gives the app a purchase token, which the app sends to our server. Our server asks Google Play (Google Play Developer API) or Apple (App Store Server API) whether that purchase is genuine and what it covers, and only then credits your account. A record of each redeemed purchase — the token stored as a one-way hash, the product, and your account ID — is kept so the same purchase can’t be redeemed twice.
  • Why: to sell and restore subscriptions and minute packs, unlock what you paid for, and prevent purchase fraud.
  • Sonary does not use a third-party subscription service. Earlier versions of this policy named RevenueCat; the app no longer includes it.

2.4 Advertising (free tier)

  • Who: ads are served by AppLovin through its AppLovin MAX SDK. Sonary currently bundles no third-party mediation adapters, so no other ad network’s SDK runs in the app. AppLovin may pass ad requests to advertisers and buyers on its own exchange, under AppLovin’s privacy policy.
  • What AppLovin collects: your device’s advertising identifier (the Google Advertising ID on Android; on iOS, the IDFA only if you allow tracking when Apple’s App Tracking Transparency prompt asks), device information such as model, operating-system version, language and network type, your IP address and the approximate location it implies, and your interactions with ads (impressions, clicks, rewarded-ad completions).
  • Your Sonary account ID: the app also gives AppLovin your Sonary account ID. When you finish a rewarded ad, AppLovin’s server sends that ID back to ours so the minute you earned is credited to the right account. It is the random ID from Section 2.2 — never your email or name.
  • Why: to show the ads that keep the free tier free (app-open, interstitial, and optional rewarded ads), and to credit rewarded-ad minutes. Ads never interrupt a job that is running.
  • Sonary Pro: Pro removes ads from the app. The AppLovin SDK is still started when the app launches, before your subscription status is known, and may load ads in the background, so it can still receive the device data above — but Pro users are not shown ads.

2.5 Analytics, diagnostics & crash reports

  • What: app-usage events — app opened, an export completed, an AI job requested and its outcome (task, Cloud or Offline tier, minutes charged, processing time) — plus events Firebase records automatically, such as first open and session start; a Firebase app-instance identifier and device/app information; crash logs and stack traces; and performance diagnostics. Our events contain no file contents, file names or paths, no text you type and no voice recordings, and we don’t attach your account ID or email to analytics or crash reports. On Android, Firebase Analytics may also read the device advertising identifier.
  • Processed by: Firebase (Google) — Analytics, Crashlytics, and Remote Config. Remote Config only delivers settings to the app (for example, a switch that can turn Cloud AI off during an outage).
  • Why: to understand which features are used, fix crashes and bugs, keep the app stable, and safely roll out configuration changes.

2.6 Data stored on your device

  • What: your Library (processing history and the files you created), recordings, your saved voice sample for Clone my voice (Section 2.8), app settings (theme, save folder, which kinds of data you allowed Cloud AI to receive), counters the app uses to space out ads and review requests, a cached copy of your minute balance, and any on-device AI models you downloaded. The Firebase Authentication SDK keeps your sign-in session in the app’s private storage (on iOS, in the Keychain).
  • Public copies: on Android, finished files are also saved to a shared folder (Download/Sonary by default — you can change the folder or turn this off in Settings). Those copies are yours and stay on your device after you uninstall Sonary. On iOS, the app’s own folder is visible in the Files app and is removed when you uninstall.
  • Notifications: the “your result is ready” notification and the one reminder about a result you haven’t saved are created and scheduled on your device. Nothing is sent to us or anyone else to deliver them.
  • Why: so your history and exports are available offline and the app remembers your choices.

2.7 Text you type for AI Voiceover

  • What: the text you type to be turned into speech (up to 5,000 characters, or 500 with a premium voice), and the voice and options you pick. The same applies to the text you type for Clone my voice (up to 500 characters; Section 2.8).
  • Where it goes: when you start a voiceover, the app sends the text over HTTPS to our backend (Firebase Cloud Functions), which checks it against a short list of clearly prohibited content (Section 2.9) and forwards it to Replicate to run the voice model: Kokoro (jaaari/kokoro-82m) for standard voices, or Chatterbox Turbo (resemble-ai/chatterbox-turbo) for premium voices. The finished audio is downloaded to your device.
  • Asked first: before your first voiceover, the app asks your permission to send typed text (“Send your text to Cloud AI?”) and names Replicate. You can withdraw it in Settings → Cloud AI data sharing → Typed text → Revoke.
  • What we keep: our backend never stores your text with the job — only if you choose to include it in a report (Section 2.9). Its record of the job keeps only the number of characters, with the task, minutes charged and status, under your account ID. Replicate keeps each job’s inputs and outputs (here, your text and the generated speech) for about 1 hour, then deletes them automatically.
  • Why: to create the speech you asked for, and to refuse a small set of clearly prohibited requests.
  • Please don’t include private information you wouldn’t want processed: the text is sent exactly as you typed it.

Clone my voice is available to Sonary Pro subscribers. It may only be used with your own voice, and it starts with your own spoken consent.

  • Your spoken consent. Our server gives you a sentence to read aloud that ends in a one-time code, for example: “I am recording my own voice, and I give Sonary permission to create a copy of it. My consent code is seven, four, two, nine, one.” You record yourself reading it in the app (5 to 30 seconds). That recording is your voice sample.
  • The check. The app uploads the recording over HTTPS to our backend, which sends it to Replicate to be transcribed by OpenAI’s Whisper speech-recognition model (openai/whisper, run on Replicate) and checks that the sentence and the code were said. The upload is deleted from Replicate right after the check, whether it passes or not.
  • The consent record. When the recording matches, our backend stores a consent record in our database (Cloud Firestore) under your account ID: a SHA-256 fingerprint of the recording (not the recording itself), the sentence, the transcript of your reading of it, the recording’s length, the time, and the App Check result (Section 2.2). The fingerprint matches only that exact file; it can’t be turned back into audio or used to recognise your voice in other recordings. We keep the consent record until you delete your account (Section 9). Our server also keeps each consent challenge it issues you (the sentence, its one-time code and the number of attempts) to enforce the limits on attempts.
  • Your saved voice. The recording that passed stays on your device, in the app’s private storage, as your saved voice. You can delete it at any time on the Clone my voice screen, or in Settings → Cloud AI data sharing → Voice sample → Revoke, which also withdraws your permission to upload it. Deleting your saved voice doesn’t delete the consent record on our server; deleting your account does.
  • Creating speech in your voice. Each time you do, the app uploads your saved voice sample and the text you typed (up to 500 characters) to our backend. The backend checks that the sample’s fingerprint matches one of your consent records, then has Replicate run the Chatterbox model (resemble-ai/chatterbox) with your sample as the voice to copy. The upload is deleted from Replicate after the job, and Replicate deletes the job’s remaining inputs and outputs (your text and the generated speech) within about an hour. The speech is downloaded to your device.
  • No voice storage on our side. Our servers never store your voice recording, and we don’t keep a voiceprint (a stored model of your voice that could identify you).
  • Biometric information. We don’t create or keep biometric identifiers. Where a law treats a voice sample as biometric information, we handle yours this way: we collect it only after your in-app permission and your spoken consent; we use it only to create the speech you ask for; we never sell, rent, trade or otherwise profit from it, or disclose it to anyone except Replicate to run your job; each upload is destroyed after its job; and the consent record (which holds no audio) is destroyed when you delete your account.
  • Asked first: before your voice sample is uploaded for the first time, the app asks your permission (“Send your voice sample to Cloud AI?”) and names Replicate. Typed text has its own permission (Section 2.7).
  • Why: to make sure a voice is only cloned by the person it belongs to, to produce the speech you ask for, and to be able to show that consent was given if a cloned voice is ever reported as misused.

2.9 Reports about generated speech

  • What: speech that Sonary generates has a Report button. A report stores your account ID, the AI tool, the job’s ID, the reason you pick (offensive, impersonation, illegal or other) and, if you add one, your note (up to 500 characters). If you tick Include the text, it also stores the text that was spoken (up to 5,000 characters), so a reviewer can see what was said; otherwise it doesn’t include your text. It never includes the audio.
  • Why: so a person can review what was reported and act on it, as Google Play’s policy on AI-generated content requires.
  • How long: until we have reviewed it, or until you delete your account if that comes first.
  • Prohibited text. Separately, our server refuses text that contains terms from a short list of clearly prohibited content, such as sexual content involving minors or explicit incitement to violence. The check runs on the text in memory; refused text isn’t stored or logged.

3. Your audio & cloud AI processing

This is the most privacy-sensitive part of the app, so we’re explicit about it.

  • Only on your action. Each AI audio tool offers two tiers: Offline AI, which runs on your device and uploads nothing, and Cloud AI, labelled “Uploads this file · Deleted after · Uses minutes”. AI Voiceover and Clone my voice run only in the cloud. Nothing is uploaded until you start a Cloud AI job yourself (or, for Clone my voice, send your consent recording to be checked). There are no background, automatic, or speculative uploads, and the app never scans your library.
  • Asked before it happens. Before the first Cloud AI upload of each kind of data (an audio file, typed text, or a voice sample), the app shows a dialog (“Send audio to Cloud AI?”, “Send your text to Cloud AI?” or “Send your voice sample to Cloud AI?”) that names Replicate, Inc. as the AI processing partner, says what is sent and what is kept, and links to this policy. Nothing is uploaded unless you tap Allow & Continue; Not now cancels the job. Your answers are stored on your device. You can withdraw each one at any time in Settings → Cloud AI data sharing → Revoke, and its dialog will appear again before the next upload of that kind. Revoking the voice sample also deletes your saved voice from the device.
  • What is sent. For an audio tool, the audio file for that one job. A file goes with its file name and any tags embedded in it (such as title or artist); files the cloud models can’t read, and audio taken from a video, are first converted on your device to a FLAC file with an app-generated name. For AI Voiceover, the text you typed and the voice and options you picked. For Clone my voice, your voice sample and, when you create speech, the text you typed. Nothing else about you is attached: no name, email, account ID, Apple ID, device identifier, advertising ID, or purchase data reaches Replicate.
  • Where it goes. Everything travels over an encrypted (HTTPS) connection to our backend (Google Cloud Functions, run through Firebase), which checks your minute balance (and, for Clone my voice, your subscription and consent record) and forwards it to our AI processing provider, Replicate, to run that one job on an AI model hosted there. Our backend records the job — task, minutes charged, status, Replicate’s job reference, links to the results, the App Check result and, for a voiceover, the number of characters (never the text) — under your account ID, so it can charge the right account and so only you can fetch the result. Replicate is never given that ID.
  • Getting the result. The app downloads audio results directly from Replicate’s file-delivery service over HTTPS. Transcripts and subtitles are served by our backend, which reads them from Replicate when you download them and doesn’t store them. As with any web server, these services see your device’s IP address for the download; no account or device identifiers are sent to Replicate with it.
  • Deletion after processing. As soon as Replicate reports the job finished — succeeded, failed, or cancelled — our backend deletes your uploaded file from Replicate through its API; a voice sample sent for the consent check is deleted right after the check. If that call ever fails, Replicate’s own automatic clean-up removes the file, and the same clean-up removes the rest of the job: the result files Replicate produced and inputs such as the text you typed (Replicate’s documentation currently gives one hour for jobs created through its API). We keep no copy of your audio, your voiceover text (unless you include it in a report, Section 2.9) or your voice sample on our own servers: uploads pass through our backend’s memory and are not stored there.
  • No training. We do not use your audio, your text or your voice to train AI models.
  • Your own voice only. Clone my voice runs only on a voice sample that passed the spoken-consent check (Section 2.8), and only for your account.
  • Offline AI stays local. Offline AI runs on-device models (htdemucs for separation, GTCRN for noise removal) that the app downloads once, after asking you, from Hugging Face and GitHub. Those downloads carry no audio; the download servers see only your IP address, as for any download.

4. What we do not collect

  • We do not require you to register. The account the app creates is anonymous; we only hold an email address or name if you choose to sign in with Google or Apple.
  • We do not store the text you type for voiceovers or your voice recordings on our servers, and we don’t keep a voiceprint.
  • We do not collect precise (GPS) location, and the app never asks for location permission.
  • We do not access your audio, media, contacts, photos, or files except the specific items you pick or record for a task.
  • We do not scan your library, upload media in the background, or read your media for advertising.
  • We do not sell your personal data, and we do not share your audio, your typed text or your voice sample with anyone except Replicate, to run a Cloud AI job you started.
  • We disclose information to authorities only when the law requires it, or when it is needed to protect someone’s safety (for example, a report or a voice-consent record connected to suspected misuse of a cloned voice).

5. Third-party processors & sub-processors

We use a small number of established providers to run parts of the service. Each handles data only for the purpose listed and under its own privacy policy.

ProviderRole in SonaryData it handlesPrivacy policy
Google Firebase — Authentication, App Check, Cloud Firestore, Cloud FunctionsOur backend: accounts and sign-in, app integrity checks, the minute ledger, purchase verification, Cloud AI job handling, voice-consent records and reportsAccount ID, linked Google/Apple sign-in details, balances, purchase and job records, App Check results, voice-consent records and reports; Cloud AI audio, typed text and voice samples in transit onlyfirebase.google.com/support/privacy
Google Firebase — Analytics, Crashlytics, Remote ConfigUsage analytics, crash reporting, remote settingsApp-instance and device identifiers, usage events, crash logs, diagnosticsfirebase.google.com/support/privacy
Google PlayAndroid billing and purchase verification, in-app updates, in-app ratings, Play Integrity (used by App Check)Purchase transactions (payment details stay with Google); the device-integrity check behind App Checkpolicies.google.com/privacy
Google Sign-In (optional)Sign-in to back up your minutesYour Google account’s email, name and account identifierpolicies.google.com/privacy
AppleApp Store billing and purchase verification; Sign in with Apple (optional); App Tracking Transparency; DeviceCheck (used by App Check)Purchase transactions; your Apple ID email or relay address and name, if you sign in; the device check behind App Checkapple.com/legal/privacy
AppLovinAdvertising (AppLovin MAX SDK)Advertising ID, device information, IP address and approximate location, ad interactions, Sonary account IDapplovin.com/privacy
ReplicateRuns the Cloud AI jobs you start (the AI audio tools, AI Voiceover, Clone my voice and its spoken-consent check) on AI models hosted there, such as OpenAI Whisper, Kokoro and Resemble AI’s ChatterboxThe audio file, typed text or voice sample for that job, and the results. Uploaded files are deleted after the job; everything else within about an hourreplicate.com/privacy
Hugging Face, GitHubHost the Offline AI model files the app downloadsYour IP address, for the download onlyhuggingface.co/privacy, GitHub privacy statement

6. How we protect your data

  • Encryption in transit. All network traffic — audio, text and voice uploads, sign-in, purchase checks, analytics, and ads — uses encrypted HTTPS (TLS) connections.
  • No secrets on the device. The Replicate API token, the store-verification keys and other provider credentials live only on our server.
  • Verified identity. Every request to our backend carries a short-lived Firebase ID token that the server verifies. A Cloud AI job’s status and results can only be read with the token of the account that started it.
  • Server-authoritative. Minute balances, purchases and subscription allowances are decided on our server from the store’s own answer, so they can’t be forged on the device.
  • Voice cloning safeguards. Clone my voice is limited to Sonary Pro, needs a fresh one-time code read aloud and checked by our server, works only with the exact recording that passed that check, and is rate-limited. Our servers never store voice recordings.
  • Minimal telemetry. Analytics events are built without file contents, file names, paths, typed text, recordings, or account identifiers.
  • No method of storage or transmission is completely secure, so we cannot guarantee absolute security.

7. Data retention

  • On-device data (Library, recordings, settings, downloaded models) stays on your device until you delete it in the app or uninstall the app. Copies saved to shared folders stay until you delete them.
  • Cloud AI audio is deleted from Replicate when the job finishes, and Replicate’s automatic clean-up removes the result files shortly after (see Section 3). We never store your audio.
  • Typed text (AI Voiceover and Clone my voice) is never stored by us, unless you include it in a report (kept until reviewed, Section 2.9). Replicate deletes it, with the generated speech, within about an hour of the job.
  • Voice samples are never stored on our servers. Each upload is deleted from Replicate right after the consent check, or after the Clone my voice job it was sent for. Your saved voice stays on your device until you delete it (Section 2.8).
  • Voice-consent records are kept until you delete your account.
  • Reports are kept until we have reviewed them, or until you delete your account if that comes first.
  • Your account and its server records — the account itself (with any linked Google or Apple sign-in), your minute ledger, Cloud AI job records, rewarded-ad records, voice-consent records and the consent challenges issued to you — are kept for as long as the account exists, and deleted when you delete the account (see Section 9).
  • Purchase records — redeemed purchase tokens, stored as one-way hashes, are kept so a purchase can never be redeemed twice. When you delete your account, your account ID is removed from them. Google Play and Apple keep their own transaction records under their policies.
  • Sign-in bonus marker — a one-way hash of each Google or Apple account that has claimed the sign-in bonus is kept permanently, so the bonus can only be claimed once per account. It contains no email address, name, or Sonary account ID.
  • Server logs — our backend’s operational logs (for example, a purchase verified or minutes credited, with the account ID) are kept by Google Cloud for a limited period under its log-retention settings.
  • Analytics, crash, and advertising data are retained by Firebase and AppLovin under their own retention policies.

8. Your rights & choices

  • Choose what leaves your device. Use Offline AI for the audio tools, or withdraw permission for each kind of data (audio files, typed text, voice sample) in Settings → Cloud AI data sharing. Withdrawing stops future uploads of that kind; it doesn’t delete records we already hold, such as a voice-consent record (deleting your account does).
  • Delete your saved voice from your device at any time (Section 2.8).
  • Notifications. You can turn Sonary’s notifications off at any time in your device settings.
  • Sign-in is optional. “Remove backup” in Settings unlinks Google or Apple sign-in at any time.
  • Delete your account from inside the app at any time — see Section 9.
  • Access & other requests. You can ask for a copy of, a correction to, or deletion of the data our server holds about your account by emailing imamagun94@gmail.com. If you signed in, include the email address of the Google or Apple account you used so we can find your account.
  • Advertising choices. On Android, you can reset or delete your advertising ID under Settings → Privacy → Ads, which also opts you out of personalised ads. On iOS, the app asks once for permission to track; you can change your answer at any time under Settings → Privacy & Security → Tracking. Declining changes nothing else in Sonary. Sonary Pro removes ads from the app.
  • Analytics. You can limit ad and analytics identifiers in your device’s privacy settings.

9. Deleting your account and data

9.1 Delete in the app

Open Settings → Account → Delete account and confirm. This is available whether or not you ever signed in with Google or Apple, and it needs an internet connection.

9.2 What is deleted

Immediately, from our servers:

  • your Firebase Authentication account, including any linked Google or Apple sign-in and the email address stored with it;
  • your minute ledger (balance, plan, allowance);
  • your Cloud AI job records (any job still running is stopped and its upload deleted);
  • your rewarded-ad records;
  • your voice-consent records and the consent challenges issued to you;
  • any reports you sent that haven’t been reviewed yet.

The app then starts over on a new anonymous account.

9.3 What is kept

  • Purchase-token records stay, with your account ID removed, so the same receipt can’t be redeemed again by a new account.
  • The sign-in bonus marker described in Section 7, which holds no email, name or account ID.
  • Store records at Google Play or Apple, under their policies.
  • Analytics and crash data held by Firebase, which isn’t linked to your Sonary account and expires under Firebase’s retention settings.
  • Advertising data held by AppLovin, including the account ID the app gave it (Section 2.4), under AppLovin’s own retention policy. After deletion the app gives AppLovin your new account ID instead. To limit advertising data, reset your advertising ID (Section 8) or contact AppLovin.
  • Server logs, until they expire (Section 7).
  • Recent Cloud AI data at Replicate (the inputs and results of jobs from roughly the last hour), until Replicate’s automatic clean-up removes it (Section 3).
  • Files on your device — Library items, exports, copies in shared folders, and your saved voice sample — until you delete them or uninstall the app.

9.4 Before you delete

  • Remaining minutes, including ones you paid for, are lost and can’t be restored.
  • Deleting your account does not cancel a subscription. Cancel it in Google Play → Subscriptions, or on iOS in Settings → [your name] → Subscriptions, or you will continue to be charged.

9.5 By email

If you can’t use the app, email imamagun94@gmail.com with the subject “Sonary data deletion request”. If you signed in with Google or Apple, send it from that account’s email address (or include it) so we can find your account. We will confirm your request and complete the deletion within 30 days.

An anonymous account that was never signed in has nothing an email can be matched to, so it can only be deleted from the app on the device that created it — use Delete account before uninstalling.

10. Device permissions we request

PermissionPlatformWhy
MicrophoneAndroid, iOSOnly to record audio when you use the voice recorder, and to record your voice sample when you read the consent sentence for Clone my voice.
Photo libraryiOSTo let you pick a video from Photos to extract audio from or process.
File access through the system pickerAndroid, iOSOnly the files you pick. The app requests no broad storage-reading permission.
Write to shared storageAndroid 8–9 onlyTo save copies of your results to Download, Music or Movies. Newer Android versions need no permission for this.
Notifications (optional)Android 13+, iOSTo tell you when a result is ready and to remind you once about a result you haven’t saved; on Android, also to show progress while an Offline AI model downloads or an Offline AI job runs. Notifications are created on your device; nothing is sent to us to deliver them.
Foreground serviceAndroidKeeps a model download or Offline AI job running if you leave the app.
Ignore battery optimizations (optional)AndroidRequested only if you turn it on in Settings, so the system doesn’t stop a long Offline AI job mid-way.
Internet and network stateAndroid, iOSFor Cloud AI jobs (including AI Voiceover and Clone my voice), sign-in, purchases, model downloads, and (free tier) ads.
Advertising IDAndroidUsed by AppLovin and Firebase Analytics as described in Sections 2.4 and 2.5.
Tracking (App Tracking Transparency)iOSAsked once, at launch, to let AppLovin use your IDFA for ads. Optional.
Google Play billingAndroidTo buy subscriptions and minute packs through Google Play.

The iOS app also contains purpose descriptions for location, camera and the media library, because code linked into the app references those system features. Sonary never asks for any of these permissions, so you won’t see those prompts.

11. Children’s privacy

Sonary is a general-purpose audio productivity tool intended for a general audience. It is not directed to children under 13 (or the equivalent minimum age in your region), and we do not knowingly collect personal information from children. If you believe a child has provided us with personal information, please contact us at imamagun94@gmail.com and we will delete it. This app is not enrolled in Google Play’s Families program and is not designed for or marketed to children.

12. International data transfers

Our backend runs on Google Cloud (Firebase) in the United States, and our providers (including Google/Firebase, Apple, AppLovin, and Replicate) may process data on servers in the United States and other countries. By using the app you understand that the limited data described here may be processed in those countries under the safeguards those providers maintain.

13. Changes to this policy

We may update this policy as the app evolves. We will change the “Last updated” date above and, for material changes, provide an in-app notice where appropriate.

14. Contact us

For any privacy question, or an access or deletion request:

imamagun94@gmail.com
NewAgeDevs — Md. Imam Hossain, Bangladesh

Related documents

Sonary: MP3 & Noise Remover AI

Site-wide

Other products (35)