1. Scope and operator
This policy applies to the Stable Audio: AI Generator iOS app, its secure backend, and its product support pages. The service is operated by Glayzer Studio under the AI Studio product identity. Apple and each AI or infrastructure provider also apply their own privacy terms to information they process.
2. No conventional account
The app does not ask you to register with a name, email address, password, phone number, social profile, or contacts. A pseudonymous service record is created from verified App Store information so purchases, credits, generation history, downloads, and deletion requests can work across sessions without a sign-in screen. We do not sell personal information. Advertising appears only when an eligible user voluntarily opens the dedicated rewarded-trial page and chooses to watch an ad.
3. Adult access
Stable Audio is intended only for people aged 18 or older. We do not request or store your birth date or exact age. The app records only that the adult entry step was completed. On supported Apple systems, a privately shared age range may be used for the access decision without being retained by Stable Audio.
4. Information processed to provide the service
- Prompts and lyrics. The text you submit, an optional corrected English version, custom lyrics, language information, and creative settings such as model, mode, duration, tempo, seed, and output format.
- Source and generated audio. MP3 or WAV files you choose, record, or import from a permitted direct link, together with generated results and the section selected for replacement. A direct-link file is downloaded and validated by your iPhone first; the link itself is removed before the generation request, while the validated audio is uploaded through the same private source-audio flow. Voice Studio keeps local mixes on your iPhone unless you explicitly choose AI analysis.
- Optional Voice Studio analysis. If you choose Analyze, the app prepares a temporary mono WAV copy of the selected performance. The service uses it to transcribe the words and detect musical delivery, tempo, genre, mood, and arrangement. The transcript and style remain editable before you choose a generation model.
- Generation history. Job status, model, provider request references, credit reservation and settlement, timestamps, output format, downloads, and the factual Generation Provenance Record.
- Pseudonymous identifiers. A protected representation of Apple’s App Transaction ID, an app installation reference, session-security data, and abuse-prevention records.
- Purchases and credits. Verified StoreKit transaction identifiers, product, entitlement, renewal or expiry state, refund or revocation state, and the append-only credit ledger needed to prevent duplicate grants and incorrect charges.
- Notifications. An Apple Push Notification service or Firebase Cloud Messaging token, device platform, app language, and notification preference so completion alerts can reach the correct installation.
- Support. If you compose an email, we receive the address, name, message, and attachments you choose to send. The website itself does not store the support form.
5. AI providers and purpose of each transfer
The production prompt and any permitted source audio are sent through our secure backend to Stability AI when you select a Stability model. For a normal 3 Clip or 3 Pro generation, Google receives the vocal production prompt and any custom lyrics; source audio is not sent for that workflow.
If you explicitly choose AI analysis in Voice Studio, Google receives the temporary mono recording to transcribe the words and analyse audible musical characteristics. Stable Audio does not ask the analysis service to identify the performer, infer protected traits, imitate a named person, or rewrite the transcript. The temporary object is deleted from our upload storage and Google Files after analysis. The resulting 3 Clip or 3 Pro song uses a new AI vocalist; it does not preserve or clone the recorded voice.
OpenAI receives prompt text only when meaning-preserving translation is required or when you choose Correct & Translate. It is instructed to correct only necessary grammar, sentence, logic, or consistency issues and not to add, remove, enrich, or reinterpret creative details. Audio is never sent to OpenAI. Requests may be rejected by provider safety systems.
To avoid charging for the same correction repeatedly and to enforce fair-use limits, the secure backend keeps a one-way fingerprint of the correction request and may reuse the corrected response for up to 24 hours. The original prompt is not stored in that cache. Expired correction responses and related token-cost records are purged by automatic maintenance, or earlier if you delete the service record. Voice-analysis request metadata used for security and rate limiting is retained for no more than 30 days and does not include the temporary recording, transcript, or detected style.
6. Optional Analytics & Diagnostics
Firebase Analytics and Crashlytics remain disabled until you enable Analytics & Diagnostics in the app. If enabled, Firebase may process safe screen names, feature interactions, app and device identifiers, network information used to infer an approximate country or region, and crash or non-fatal diagnostic details. Stable Audio does not send prompts, lyrics, audio, filenames, precise location, advertising identifiers, or its service-account identifier to Firebase Analytics. You can disable collection again in Settings.
7. Optional rewarded advertising
In eligible countries, you may voluntarily watch a Google AdMob rewarded ad on a dedicated trial page to receive the displayed Music Credits for one Lyria 3 Clip generation. Ads are not shown on the home, create, library, player, Premium, onboarding, or Settings screens.
Google's User Messaging Platform consent flow and Apple's App Tracking Transparency choice are requested only when you open that page. You may refuse and leave the page without losing purchased access. Where required, the page also provides advertising privacy controls. Depending on your choices, Google may process an IP-address-derived approximate location, device or advertising identifiers, advertising data, product interactions, performance information, crash logs, and other diagnostic data to deliver, limit, secure, measure, or personalise the ad.
Stable Audio records the pseudonymous service owner, server-observed country, reward date, test or production ad-unit reference, and granted credits to enforce the displayed daily trial limit and prevent abuse.
8. Infrastructure and other recipients
Cloudflare processes encrypted service traffic and stores private audio objects needed for uploads, results, and exports. Apple processes App Store purchases, subscription management, App Attest, and push delivery. Firebase provides optional analytics and crash diagnostics as described above and may assist with notification-token delivery. Google AdMob and its User Messaging Platform process advertising and consent information only for the dedicated rewarded-trial flow described above. Stability AI, Google, and OpenAI process only the request content needed for the model or text feature you explicitly select. These services receive only the information needed for their role.
9. Why information is processed
Information is processed to perform the operation you request, translate or correct a prompt when authorised, deliver and preserve results, verify purchases, maintain credits, send enabled notifications, provide an optional rewarded ad and apply its credit reward, prevent fraud and abuse, investigate failures, respond to support, comply with law, and protect the service. Where applicable, these purposes rely on performing the service contract, your consent, legitimate security and reliability interests, or a legal obligation.
10. Storage and retention
Prompts, settings, uploaded or recorded audio, generated audio, history, and provenance data may remain associated with the pseudonymous service record so you can return to and download prior creations. You can delete settled creations, clear completed history, export your account record, or request complete account deletion in Settings.
A recording selected for Voice Studio AI analysis is limited to five minutes and 30 MB after preparation. Its temporary upload is removed after analysis whether the analysis succeeds or fails. If deletion cannot be confirmed, the request fails rather than proceeding as if cleanup succeeded. Abandoned, unreferenced input uploads are also removed automatically.
Active jobs and jobs awaiting an uncertain provider-credit settlement are retained until they can be reconciled safely. Account deletion immediately prevents new use and completes after that reconciliation. Purchase and ledger evidence may be retained as required for refunds, tax, disputes, fraud prevention, or legal compliance. After account deletion, keyed transaction fingerprints that cannot identify an owner may remain solely to prevent duplicate consumable-credit grants. Support correspondence is kept only as long as reasonably needed to answer and document the request.
11. Security
Provider credentials are kept in encrypted server-side secrets and are not included in the iOS app. Data is transmitted over encrypted connections, audio storage is private, and access is protected with Apple-backed session verification and platform security controls. No system can be guaranteed completely secure, but safeguards are selected for the sensitivity of the information processed.
12. Your controls and rights
Settings lets you withdraw future AI-processing permission, disable optional analytics, change notification preferences, export your service record, delete completed history, or permanently delete the pseudonymous account and associated service and device data. Account deletion does not cancel an Apple subscription; subscriptions must be managed separately in the App Store. You may also contact us about access, correction, deletion, restriction, objection, portability, or consent rights that apply in your region. Advertising privacy choices can be reopened from the rewarded-trial page where required, and Apple tracking permission can also be changed in iOS Settings.
13. International processing and changes
Service providers may process information in countries other than your own using contractual and legal safeguards where required. We will update this policy when providers, product behaviour, or legal requirements materially change and will present renewed consent in the app when necessary.
14. Contact
Send privacy or deletion questions through the Stable Audio support page. Do not include provider API keys, payment-card details, or source audio that is not necessary for the request.
Contact developer