Whisper Terminal 0.1.0 for Windows (internal preview)
==========================================================

Dictation anywhere, on your graphics card, and terminals that open over whatever you are doing.
This is the first Windows build. It was built on a Linux machine and has not yet been run on a real
Windows PC by the people who made it: you are the first test. Please tell us what you see.

Needs: Windows 10 22H2 or Windows 11, 64-bit. A microphone. For the graphics card: a current graphics
driver from NVIDIA, AMD or Intel (the driver brings the Vulkan library the speech helper uses).
Nothing else: no Visual C++ Redistributable, no administrator rights.

Start it
--------
Unpack the zip anywhere (a folder you own, not "Program Files") and double-click whisper-terminal.exe,
or run the installer (WhisperTerminal-0.1.0-windows-x86_64-setup.exe: per user, no administrator).
A small icon appears in the notification area (click ^ if Windows hides it) and the terminals open once (that is what
starting it does; the "start when I sign in" option starts it quietly, with only the icon).

* Windows may say "Windows protected your PC" (SmartScreen). The program is not signed yet, so every new
  program from a small publisher gets that. Choose "More info", then "Run anyway". The SHA-256 of what
  you downloaded is in SHA256SUMS: in PowerShell, `Get-FileHash <file>` shows yours.
* Windows Defender may take a few seconds to look at the first start.

What works in this build
------------------------
* Dictation:    press Ctrl+Space, speak, press Ctrl+Space again. The text is pasted into the window you
                were in (it is on the clipboard too). The first dictation downloads a speech model (the
                size depends on your PC; a progress line shows in Settings).
* Your GPU:     Settings > Speech says which device runs it, in words ("Speech: GPU (Vulkan, NVIDIA ...)"),
                and every dictation writes a line to the log with the device and the time it took.
                If it says CPU although you have a graphics card, it tells you why and what to do.
* Terminals:    press Win+Enter. Git Bash opens (PowerShell or Command Prompt if Git is not installed),
                with tabs and splits. `claude` runs in it. Ctrl+Shift+Space dictates into the terminal,
                Ctrl+Shift+Enter dictates and presses Enter.
* Settings, History: from the tray icon's menu. The tray icon: left click opens the terminals (or
                stops a recording), right click is the menu.

What does not work yet (Linux only for now)
-------------------------------------------
Computer use (the agent that clicks and types for you), the connector hub (Telegram, Slack, the web page,
the phone), WhisperKey import, the sounds, and the DIY OS parts. The window for them is not in the menu.
Pasting into a program that runs "as administrator" is refused by Windows; the text stays on the
clipboard and the orb says "copied".

A ten-minute test (what we most want to know)
---------------------------------------------
 1. Start it. Does an icon appear in the notification area, and do the terminals open (Git Bash)?
 2. Right-click the icon: a menu with Start dictation, Terminals, History, Settings, Quit?
 3. Settings > Shortcuts: do the shortcuts say "registered"? If one says Windows refused it, another
    program uses that key: tell us which key (you can change it in settings.toml, then press "Rebind").
 4. Settings > Speech: what does the first line say? "Speech: GPU (Vulkan, <your card>)" is what we hope for.
 5. Open Notepad, press Ctrl+Space, say a sentence, press Ctrl+Space again. The first time a speech model
    is downloaded (a minute or two). Does the text appear in Notepad? How long after the second press?
 6. The same in a browser's text field, and in a program that runs as administrator (it should say "copied").
 7. Win+Enter: the terminal. Does `claude` run in it? Does Ctrl+Shift+Space dictate into it?
 8. The report and the log file (below): copy both. In PowerShell, from the folder:
        .\whisper-terminal.exe --doctor | Out-File doctor.txt
    (in cmd: whisper-terminal.exe --doctor > doctor.txt, and wait a second before opening doctor.txt: cmd does not wait for it.)
 9. In the terminal (Win+Enter) type  sleep 300 &  then close that tab, and Quit from the menu. Does the icon go?
    Are whisper-terminal.exe, wt-stt.exe, bash.exe and sleep.exe all gone from Task Manager (Details tab)?
10. Anything that looked wrong, with a screenshot if you can.
The tests zip (WhisperTerminal-0.1.0-windows-x86_64-tests.zip, if you were given it) holds the program's own tests
as .exe files: unpack it, run run-tests.cmd, and send what it prints (a few failures are expected on a first run:
they are the first native run of tests that were only ever compiled for Windows).

If something does not work
--------------------------
1. whisper-terminal.exe --doctor   prints what this PC looks like to it. Save it to a file, as in step 8 of the test above
   (.\whisper-terminal.exe --doctor | Out-File doctor.txt in PowerShell).
2. The log:  %LOCALAPPDATA%\whisper-terminal\whisper-terminal.log   (and wt-stt.log beside it for the speech helper)
3. Send both to whoever gave you the zip, with what you pressed and what you expected.
   Useful too: whisper-terminal.exe --status   and   whisper-terminal.exe --stt probe

Your data
---------
Settings:  %APPDATA%\whisper-terminal\settings.toml  (shortcuts are edited there; Settings shows which ones
           Windows accepted, and which another program already has)
History, recordings, models, log:  %LOCALAPPDATA%\whisper-terminal\
Nothing is sent anywhere except the one model download (from Hugging Face, with the curl that comes with Windows).
To remove it: quit from the tray menu and delete the folder (or run the uninstaller), then the two folders above.
