Data Storage

Everything Speakly keeps sits in one folder on your own machine: the app's application data folder. Nothing on this page is stored on our servers.

Where the folder is

• macOS: ~/Library/Application Support/Speakly
• Windows: %APPDATA%\Speakly

What is inside it

• speakly-config.json — settings, preferences, API keys, and your dictation history (the most recent 1000 entries).
• whisper-models/ — downloaded local Whisper models, one .bin file per model.
• audio/ — saved recordings, only when Keep audio recordings is on. Files are named audio_<timestamp>_<random>.webm.
• IndexedDB/ — Dictionary, Notes, Custom Prompts, and the transcripts of uploaded files, YouTube videos and live sessions. The database is speakly-offline-<your user id>, so two accounts on one machine never share data.
• llm-models/ — downloaded local LLM models used by post-processing.
• Cache/, Code Cache/, GPUCache/ — Chromium caches, rebuilt automatically.

Seeing where the space went

Settings → Statistics measures this folder without leaving the app: it prints the total as "… in Speakly's data folder" and splits it into Whisper models, Local LLM models, Saved audio files, History and settings, Caches and Everything else. Rows measuring zero are hidden. See Statistics.

• Whisper models — the app's hint is "Delete unused sizes in Transcription"
• Local LLM models — "Delete unused sizes in Post-Processing"
• Saved audio files — "Cleared by “Delete history” in Data and Privacy"
• History and settings — speakly-config.json plus IndexedDB/
• Caches — "Rebuilt automatically"

Is any of it encrypted?

speakly-config.json is written encrypted. The key is derived at runtime from your own application data path, so it is not in the source code and the file cannot simply be opened in a text editor. That also means the file does not travel: copied to another machine it no longer decrypts, and Speakly starts a fresh one.

The IndexedDB data — dictionary, notes, custom prompts, and file, YouTube and live transcripts — is not encrypted, and neither are saved audio files or models. If you need protection at rest for those, turn on FileVault on macOS or BitLocker on Windows.

Where saved audio goes

Audio is deleted the moment a transcription finishes unless you change that. The toggle is Keep audio recordings in Settings → System, and it states its own behaviour: off it reads "Audio files are deleted immediately after transcription.", on it reads "Audio files are saved locally. Delete them with transcriptions in Recent Activity." Kept files land in the audio/ folder above and are removed together with the transcription they belong to. Keeping the audio is also what makes Re-transcribe with the current model possible on a past dictation in Recent Activity: with the toggle off there is no file left to run again.

How to delete everything

1. Settings → Data and Privacy → Delete history clears the dictation history under Recent Activity and the audio files kept with it. The confirmation is Delete transcription history? and its button is Delete History.
2. Settings → Data and Privacy → Erase & Restart clears the rest and restarts the app: API keys, that same history and audio, Dictionary, Notes and Custom Prompts, app settings and preferences, and the login session. Your account on the server is untouched.
3. Downloaded models survive both. Remove them one at a time in Settings → Transcription — trash icon, then confirm Delete Model? — or delete whisper-models/ in the folder above.