|
1 | 1 | # Blitztext Linux Roadmap |
2 | 2 |
|
3 | | -This roadmap describes the Linux-focused direction of the repository. |
| 3 | +This roadmap tracks open work and near-term ideas for the Linux repo. |
4 | 4 | It is a planning note, not a promise. |
5 | 5 |
|
6 | 6 | ## Current scope |
7 | 7 |
|
8 | 8 | - Linux desktop app for Ubuntu/Kubuntu on KDE Plasma and Wayland |
9 | 9 | - PyQt6 tray application with a real window fallback |
10 | | -- Global hotkeys via `evdev` |
| 10 | +- Global hotkeys via `evdev` (toggle and hold) |
11 | 11 | - Local transcription via `openai-whisper` and optional `faster-whisper` |
12 | 12 | - Optional OpenAI rewriting workflows |
13 | | -- Diktat, Verlauf, Vorlesen, and notifications |
| 13 | +- Diktat, Verlauf, Vorlesen, Audio-Export, and notifications |
14 | 14 | - Install / verify / autostart flow driven by `scripts/install.sh` and `scripts/verify.sh` |
15 | 15 |
|
16 | 16 | ## Next useful work |
17 | 17 |
|
| 18 | +### Docs / repo hygiene |
| 19 | + |
18 | 20 | - Keep the Linux README and `docs/` tree aligned with the actual app behavior |
19 | 21 | - Tighten CI coverage for repo-root docs and scripts |
20 | 22 | - Reduce stale local artifacts and improve repo hygiene |
21 | | -- Improve the X11 fallback story or document the Wayland-only limitations more clearly |
22 | 23 | - Keep the installer and verify script in sync with the dependencies they check |
| 24 | + |
| 25 | +### Core product work |
| 26 | + |
23 | 27 | - Add more regression coverage around startup, config, and transcription edge cases |
| 28 | +- Improve the X11 fallback story or document the Wayland-only limitations more clearly |
24 | 29 | - Replace the current `evdev`/`input` global-hotkey path with a desktop-native XDG GlobalShortcuts integration when it is practical for KDE/Wayland |
25 | 30 | - Add a lightweight launch smoke test (boot the app offscreen and exit cleanly) so CI confirms the GUI actually starts, not just that mocked logic passes |
26 | | -- Export transcribed text as a shareable audio file (e.g. OGG/Opus or MP3) so longer transcripts can be sent as a voice message via WhatsApp or similar (builds on the TTS pipeline) |
| 31 | + |
| 32 | +### Board ideas worth keeping |
| 33 | + |
| 34 | +- Harden `PasteService` with clipboard restore and a terminal-paste fallback |
| 35 | +- Research streaming / VAD to lower dictation latency |
| 36 | +- Research IBus / input-method integration as a long-term successor to the current paste flow |
| 37 | +- Add a context mode that can suggest the right preset or tone before rewriting |
| 38 | + |
| 39 | +## Already shipped / not open roadmap items |
| 40 | + |
| 41 | +- Audio export from Read Aloud is already implemented |
| 42 | +- Hold-to-talk is already available via the existing hotkey mode |
27 | 43 |
|
28 | 44 | ## Paket H — Audit hardening (2026-06-21) |
29 | 45 |
|
|
0 commit comments