Feature Request: Add Optional "Cloud Dictation" Mode*

[wrn491]

Hi LeanBitLab Team,

First off, love LeanType! The offline STT is fantastic for privacy. But there are 2 major challenges right now:

Current Issues:

1. Offline models are very large and take up a lot of phone storage.
2. Accuracy and language support is limited compared to cloud models.

Feature Request:
Add an optional "Cloud Dictation" mode in Settings.

Proposed Implementation:

1. Default: Keep Offline STT as default

2. New Setting: Dictation Provider
   Let users choose their preferred provider:

   • Offline
   • Duck.ai API - No account required, privacy-focused
   • OpenAI Whisper API - Use with API Key
   • HuggingFace Inference API - Open-source option

3. Privacy First: Show the respective API's privacy notice before enabling. Ex: "Audio is not stored and not used for training"

Benefits:

1. No large model downloads required
2. Much better accuracy + support for 90+ languages
3. User choice: Users can balance privacy vs accuracy vs cost themselves

This would make LeanType usable for way more users without forcing huge downloads.

Thanks for considering!

---