Windows install

Two paths. WSL2 is the recommended one for full feature parity (including Korean MeloTTS); native Windows works with a couple of caveats.

Path 1: WSL2 (recommended)

The Linux installer just works inside WSL2 — the existing install.sh runs as-is.

1. Install WSL2 if you don't have it:

   wsl --install

   This installs Ubuntu by default. Reboot if prompted.

2. Open the Ubuntu shell and install prereqs:

   sudo apt-get update
   sudo apt-get install -y git curl espeak-ng python3-venv build-essential python3-dev nodejs npm
   # If you want Korean / MeloTTS:
   sudo apt-get install -y libmecab-dev mecab-ipadic-utf8 libssl-dev pkg-config

3. Install Node 20+ (Ubuntu's default is too old):

   curl -fsSL https://deb.nodesource.com/setup_20.x | sudo -E bash -
   sudo apt-get install -y nodejs

4. Run the one-liner as you would on Linux:

   bash <(curl -fsSL https://raw.githubusercontent.com/powder-nomad/papercup/main/install.sh)

5. Discord voice from WSL2 — works fine. WSL2 has full network stack access.

Path 2: Native Windows (PowerShell)

For when WSL2 isn't an option. Caveats:

• MeloTTS is not available on native Windows — its MeCab + libssl deps don't ship cleanly. The PowerShell installer auto-switches to kokoro if you ask for melotts/auto.
• Korean TTS still works via -Tts xtts (Coqui XTTS-v2 has Windows wheels). XTTS gives you ~58 built-in speakers + voice cloning from a 6-second clip.

Prereqs

Install these manually:

Package	Where
Git for Windows	https://git-scm.com/download/win
Node.js LTS (20+)	https://nodejs.org/ — pick the LTS Windows installer
Python 3.10+	https://www.python.org/downloads/windows/ — check "Add to PATH" during install
espeak-ng	https://github.com/espeak-ng/espeak-ng/releases — run installer + add to PATH
Visual Studio Build Tools	https://visualstudio.microsoft.com/visual-cpp-build-tools/ — needed for onnxruntime-node native deps if pre-built fails

One-liner installer

In an Administrator PowerShell (right-click → Run as Administrator):

iwr -useb https://raw.githubusercontent.com/powder-nomad/papercup/main/install.ps1 | iex

If your execution policy blocks the IEX, allow it for this session:

Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass
iwr -useb https://raw.githubusercontent.com/powder-nomad/papercup/main/install.ps1 | iex

With flags

# Download once, run with custom flags
iwr -OutFile install.ps1 https://raw.githubusercontent.com/powder-nomad/papercup/main/install.ps1
powershell -File install.ps1 -Agent claude-code -Tts xtts -Voice af_heart

All flags (mirror of install.sh):

-Dir <path>             Install location. Default: %USERPROFILE%\papercup
-Branch <name>          Git branch. Default: main
-DiscordToken <token>   Bot token (otherwise prompted)
-DiscordClientId <id>
-DiscordGuildId <id>

-Agent <name>           claude-code (default) | codex | anthropic-api
-Model <name>           haiku (default), sonnet, opus, gpt-5, etc.
-AnthropicApiKey <k>    Required only if -Agent anthropic-api

-Vad <name>             silero (default; only option today)
-Stt <name>             whisper-base (default) | whisper-small |
                        whisper-base.en | whisper-small.en
-Tts <name>             kokoro (default) | xtts (Korean + voice cloning)
-Voice <name>           Kokoro voice. Default: af_heart

-SilenceMs <int>        End-of-utterance silence (ms). Default: 600
-VadThreshold <float>   Speech probability threshold. Default: 0.4

-SkipModels             Don't download voice models
-SkipVenv               Don't create the Python venv
-SkipRegister           Don't push slash commands to Discord
-NoStart                Install only; don't start the daemon
-Yes                    Accept all defaults; never prompt

Daemon control

After install, manage the bot from PowerShell:

# All commands run from anywhere
$papercup = "$env:USERPROFILE\papercup\packages\bot\bin\papercup.ps1"

powershell -File $papercup status
powershell -File $papercup start
powershell -File $papercup stop
powershell -File $papercup restart
powershell -File $papercup logs    # Get-Content -Wait
powershell -File $papercup tail 50

Or cd %USERPROFILE%\papercup\packages\bot\bin and call .\papercup.ps1 <command> directly.

Common Windows gotchas

• onnxruntime-node native dep fails on npm install — install Visual Studio Build Tools (link above) and re-run the installer.
• espeak-ng not on PATH — run the installer, then either reboot or run setx PATH "$env:PATH;C:\Program Files\eSpeak NG" and open a new shell.
• "running scripts is disabled" on PowerShell — Set-ExecutionPolicy -Scope CurrentUser RemoteSigned (one-time, persistent).
• Discord voice cuts out / no audio — Windows defaults often add audio enhancements that mess with packets. In Sound Settings → Device properties → Additional device properties → Enhancements → check "Disable all enhancements" for the input mic.
• Bot silently dies on startup — check %USERPROFILE%\papercup\packages\bot\logs\bot.log.err (PowerShell Start-Process writes stderr to a separate file).