1. How We Handle Information
This app is designed for local GGUF model management, local inference, and local speech-to-text transcription. Model files, chat records, system prompts, generation settings, runtime logs, preferences, model-folder references, recorded or imported audio/video files, transcripts, transcription history, timestamped segments, sentence/paragraph mappings, and playback state are stored on the user's device by default.
If future versions offer a free download with a Pro in-app purchase, the app may store the free message usage count, purchase verification cache, entitlement status, and restore-purchase state locally on device. This is used to decide whether to show Pro entry points, whether to allow local features that require Pro, and whether to preserve confirmed access while online verification is temporarily unavailable. Purchases, refunds, billing, full payment credentials, and Apple Account information are handled by Apple. The developer does not directly receive user bank card numbers, payment accounts, or full purchase credentials.
2. Accounts and Cloud Inference
The current version does not require an account, does not provide a developer-operated cloud model inference or speech-recognition service, and does not actively upload user prompts, chats, generated results, recordings, imported audio, imported video, or transcripts to developer servers because of local inference or local speech-to-text features.
3. Legal Documents and Developer Website Access
The app may load public legal documents from the developer website through an in-app WebView or system browser, such as the Privacy Policy, Terms of Use, Third-Party Model Statement, Open Source Licenses, and Disclaimer. When these pages are accessed, the hosting provider may process IP address, request time, visited URL, browser, or client request information under its own service operations. This access is only for displaying legal documents or public notices; it is not cloud model inference and is not used to upload user chats, prompts, model files, or generated results.
4. Folder Authorization and Local Storage
The model library folder selected by the user is authorized through the system file picker. The app stores necessary security-scoped bookmarks or folder references, model names, sizes, download states, download progress, and local catalog information to show model lists, load GGUF files, manage downloads, and restore the workspace.
When the user records audio, the app requests microphone permission and saves the recording inside the app sandbox for local transcription, history, and playback. When the user imports audio, video, or a folder containing media files, the app reads the files selected through the system file picker and copies supported media files into the app sandbox for local transcription and history playback. Transcription history may store raw segments, formatted transcript text, punctuation and paragraphing results, sentence-level or paragraph-level timeline mappings, audio file references, duration, model name, language information, and creation time.
5. Network Access
Network access mainly occurs when the user actively downloads recommended chat models or Whisper speech models, visits model source pages, opens external links, loads legal documents from the developer website in the app, or when future versions add online features. Third-party platforms or website hosting providers may process IP address, request time, download URL, visited URL, and client request information under their own privacy policies.
6. Third-Party Models and Platforms
Third-party chat models, Whisper speech models, platforms, and download sources may have their own privacy policies, terms, licenses, and usage restrictions. Before downloading, importing, or running a model, users should confirm that the source is trustworthy, the file is complete, the license permits their use, and the model is suitable for their purpose.
Whisper speech model entries are used to obtain static model weight files required for local speech recognition. Speech recognition runs locally on device; third-party download platforms may only process network request information generated when the user actively downloads a model.
7. Local Records, Deletion, and Backups
Users can delete model records, chat records, system prompts, and transcription history in the app, and can manage the authorized model folder in the system Files app. Deleting a transcription history item usually removes the corresponding record; if a related audio file is stored in the app sandbox, the app attempts to clean up the related reference or file. Deleting the app removes app data according to iOS / iPadOS rules. System backups or iCloud backups may separately include some local data depending on Apple system behavior and user settings.
8. Diagnostic Information
The app may show runtime logs, device technical information, model loading status, audio transcription status, audio-reading errors, Whisper model errors, and other error status on device. Natural-language log diagnostics are generated on device by the currently loaded local model and are for reference only.
9. System Media Controls and Background Playback
When the user plays a recording, imported audio, or historical transcription audio inside the app, the app may provide the current playback title, playback position, duration, play/pause state, and app icon to iOS / iPadOS system media controls so the user can pause or resume playback from Control Center, the Lock Screen, Dynamic Island, or other system playback surfaces. This information is used for system-level playback control and does not mean that developer servers receive the audio content or transcript text.
10. Future Feature Changes
If future versions add analytics, subscriptions, cloud sync, online model services, third-party SDKs, or other online capabilities, this Privacy Policy and the App Store privacy information will be updated based on the actual features.
11. Data Collection, Use, Sharing, and Deletion Summary
The current version does not collect account information that identifies the user, does not require account registration, and does not use developer-operated servers to store user chats, prompts, model files, generated results, recordings, imported audio, imported video, or transcripts.
The app stores the following data locally on device: chat records, system prompts, generation settings, runtime logs, model-folder references, model names, model sizes, download states, download progress, preferences, recording files, imported media files, transcription history, raw transcription segments, formatted transcript text, sentence/paragraph timeline mappings, playback state, and some app state data. This data is used to provide local chat, model loading, model management, download restoration, local speech-to-text, history playback, text-to-audio seeking, playback highlighting, error diagnostics, and interface state restoration.
The app does not sell local chats, prompts, generated results, model files, recordings, imported media, transcripts, runtime logs, or transcription history to third parties, and does not use them for advertising tracking.
When the user actively taps a third-party model download entry, visits a model source, opens an external link, or visits a developer website legal page, the request may be sent to the corresponding third-party platform or website hosting provider. That platform or provider may process IP address, request time, download address, visited URL, and client request information under its own privacy policy. This app does not control third-party platform or hosting-provider data handling.
Users can delete chat records, system prompts, model records, transcription history, and related local data in the app, or delete the app through iOS / iPadOS settings to remove data in the app sandbox. Users can also delete or move the model library folder in the system Files app. Data deletion and restoration in system backups or iCloud backups are controlled by Apple system behavior and user settings.
12. Contact
For questions about this Privacy Policy or data handling, contact: 2891423931@qq.com.