Releases: cauyxy/sotoapp
Releases · cauyxy/sotoapp
Release list
v0.2.1: Soto Electron Refactor
Soto 0.2.1
Hold a hotkey, speak, release — your transcript lands wherever your cursor is.
0.2 is a ground-up rewrite. Soto moves from the original Rust/Tauri app to an Electron + React + TypeScript core — the same hold-to-talk flow, now with a refined capsule, a new Models cockpit, user-creatable modes, and more reliable text injection on macOS and Windows.
Highlights
- Rebuilt on Electron + React + TypeScript — a full rewrite of 0.1.0; identical hold-to-talk UX, cleaner foundation
- Refined voice capsule — a clearer lifecycle (idle → listening → polishing → inserted) and a distinct Translate mode (文 ⇄ A); other audio auto-mutes while you record, with subtle start/stop cues
- Models cockpit — one page to add, verify, and switch providers, with brand logos, prettified model names, and "get API key" shortcuts
- User-creatable modes — create, rename, and delete your own dictation modes (name + prompt + hotkey), alongside Default and Translate
- More providers — MiMo, Doubao, and Qwen out of the box, plus any OpenAI-compatible endpoint. Recommended:
doubao-seed-2-0-lite-260428or Qwen3.5-Omni Flash - Clipboard-safe injection — paste-based insertion now captures and restores every clipboard format, so delivering your transcript never clobbers what you had copied
- Launch at login — optional auto-start, toggled in Settings
Downloads
| Platform | File |
|---|---|
| macOS (Apple Silicon) | Soto_0.2.1_aarch64.dmg |
| Windows (x64) | Soto_0.2.1_x64-setup.exe |
Upgrading from 0.1.0
This is a fresh build on a new on-device data schema — settings and history from 0.1.0 don't carry over. After installing, re-add your provider on the Models page and re-grant macOS Microphone + Accessibility permissions.
Known limitations
- macOS Intel not supported (Apple Silicon only)
- Context signals (focused window, clipboard hints) planned, not in this release
- Partial streaming transcript not yet available
Soto 0.1.0
Soto 0.1.0
Hold a hotkey, speak, release — transcript injected wherever your cursor is. First public release for macOS and Windows.
Highlights
- Voice capsule — floating indicator that confirms the mic is live and shows session state
- Three AI providers — Mimo, Doubao Ark, and Qianwen (DashScope) out of the box. Recommended:
qwen3.5-omni-plusordoubao-seed-2-0-lite-260428 - Modes — Default, Translate, and fully customisable prompts
- Dictionary — hot words passed to the provider on every session
- Text injection — native keystrokes on Windows, clipboard paste on macOS
- History — searchable local session log
Downloads
| Platform | File |
|---|---|
| macOS (Apple Silicon) | Soto_0.1.0_aarch64.dmg |
| Windows (x64) | Soto_0.1.0_x64-setup.exe |
Known limitations
- macOS Intel not yet supported
- Context signals (focused window, clipboard) planned but not in this release
- Partial streaming transcript not yet available