VoiceLaunch TTS

A free, open-source text-to-speech launcher that turns your typing into a real microphone for Discord and any game.

Type a phrase → hear a natural voice → it comes out of your virtual microphone so everyone in your Discord call or game voice chat hears it. Hundreds of Microsoft Edge TTS voices work instantly with no install and no account. When you go offline, a local voice takes over automatically. No telemetry.

Built for people who can't or prefer not to speak, for gamers who want a voice in chat, and for anyone who needs fast, private, assistive communication.

🌐 Website: skarl007.github.io/sound_voice · ⬇️ Download

---

Why VoiceLaunch

• Your voice in Discord and any game. The generated speech is routed into a virtual microphone (VB-Cable), so any app that reads from a mic — Discord, Zoom, in-game voice chat — hears it. VB-Cable ships inside the installer and sets itself up.
• Smart voice routing. In Auto mode the app uses online Edge TTS when you have internet and instantly falls back to a local voice (Kokoro/Piper) when you're offline or the connection drops — no setting to flip.
• Instant, no account. Hundreds of natural Edge TTS voices work the moment you open the app. Nothing to download, no sign-up, no telemetry.
• Global voice shortcuts. Bind a phrase to a hotkey and fire it from inside a game without alt-tabbing.
• Built for accessibility. Persistent history and draft, quick phrases, an on-screen keyboard, high-contrast and large-font modes, full keyboard navigation, and visible focus.

---

Install

Download (recommended)

1. Go to Releases.
2. Download the latest installer (VoiceLaunch-TTS-Setup-<version>.exe).
3. Run it. The bundled VB-Audio Virtual Cable installs automatically if it isn't already present (a Windows restart may be required).

Windows SmartScreen

This is an unsigned open-source build, so Windows shows "Windows protected your PC." Click More info → Run anyway. This is expected for unsigned installers.

Verify the download (optional but recommended)

Every release publishes the SHA-256 of the installer. Confirm your download matches:

Get-FileHash -Algorithm SHA256 .\VoiceLaunch-TTS-Setup-<version>.exe

Compare the output against the hash in the release notes.

System requirements

Requirement	Minimum	Recommended
OS	Windows 10 x64	Windows 11 x64
RAM	4 GB	8 GB
Storage	2 GB free	10 GB (for local models)
GPU	Any (Edge TTS runs on all hardware)	NVIDIA with CUDA (for XTTS v2 cloning)
Internet	Required for cloud Edge TTS voices	Optional once local models are downloaded

---

Use it in Discord (or any game)

1. Open VoiceLaunch and enable the virtual mic (the app auto-selects CABLE Input).
2. In Discord: Settings → Voice & Video → Input Device → CABLE Output (VB-Audio Virtual Cable).
3. Type a phrase and hit Speak, or fire a voice shortcut hotkey while the game has focus.

Anyone in the channel now hears your generated voice. The same CABLE Output device works as the "microphone" in any game's voice-chat settings.

---

Development

Prerequisites

• Windows 10/11
• Node.js 20+
• Python 3.12 (only needed to build the bundled local backend)

Quick start

git clone https://github.com/skarL007/sound_voice.git
cd sound_voice
npm install
npm run dev

The app opens with cloud Edge TTS voices working immediately. Local voices (Piper/Kokoro) require the packaged Python backend — see CONTRIBUTING.md.

Verify

npm run type-check
npm test

Build the installer

npm run dist:win

For architecture and how the pieces fit together, see docs/ARCHITECTURE.md.

---

How it works

flowchart LR
    A[Type text or fire a shortcut] --> B[VoiceLaunch]
    B --> C{Auto routing}
    C -->|online| D[Edge TTS - cloud]
    C -->|offline / Edge down| E[Local voice - Kokoro/Piper]
    D --> F[Audio]
    E --> F
    F --> G[You hear it - monitor]
    F --> H[CABLE Input - virtual mic]
    H --> I[Discord, Zoom, games]

The cloud path renders in the app itself and fans the audio out to the virtual-mic device; the local path renders in the Python backend. See docs/ARCHITECTURE.md for the full picture.

Voices

Model	Status	Notes
Edge TTS (cloud)	Default	Hundreds of voices, no install, needs internet
Piper	Stable (local)	Lightweight, CPU-only, works offline
Kokoro	Stable (local)	Higher quality (MOS 4.2), still CPU-friendly
XTTS v2	Advanced (local)	Voice cloning; NVIDIA + CUDA only
MeloTTS / Fish Speech / Bark	Experimental	Outside the main flow

Local voices are auto-selected by hardware tier in Auto mode. The recommended local path is Piper first, Kokoro for better quality.

Repository layout

src/
  main/       Electron main process (lifecycle, IPC, Python backend, Edge TTS client)
  preload/    secure context bridge to the renderer
  renderer/   React app (Speak, Shortcuts, Settings, compact mode)
  python/     FastAPI backend and TTS engine wrappers
  shared/     shared TypeScript types
docs/         architecture, accessibility, virtual mic, release validation

Documentation

• Architecture
• Accessibility
• Virtual microphone guide
• Release validation protocol
• Contributing
• Security policy

Data & logs

• Logs: %APPDATA%\voicelaunch-tts\logs\
• Models: %APPDATA%\voicelaunch-tts\models\
• Cloned voices: %APPDATA%\voicelaunch-tts\voices\

License

MIT — see LICENSE.