Summary

Voquill provides an open-source, cross-platform speech-to-text workspace enabling users to dictate into any desktop application, aiming for 4x typing speed. It addresses input speed limitations via seamless voice dictation, AI transcript cleanup, and a synchronized personal glossary, targeting power users and developers.

How It Works

The project uses a Turborepo monorepo with a Tauri desktop app (React/Zustand) and Rust API layer for native functions. Users choose between local Whisper STT (CPU/GPU) or Groq's hosted Whisper via API key. An AI pipeline (@repo/voice-ai) automatically refines transcripts. A personal glossary ensures term accuracy. Firebase Functions handle auth and billing.

Quick Start & Requirements

• Installation: npm install after cloning.
• Prerequisites: Node.js 18+, npm 10+, Rust toolchain (incl. Tauri CLI), platform dependencies (Linux/Windows scripts provided). Groq API key (hosted STT/cleanup), Firebase CLI (backend).
• Build & Run: npm run build for full build. Desktop app via platform scripts (e.g., npm run dev:linux:gpu). Marketing site: npm run dev --workspace apps/web.
• Resources: Website: https://voquill.com. Discord: https://discord.gg/5jXkDvdVdt.

Highlighted Details

• System-wide voice input (overlay, hotkeys, integrations).
• Flexible STT: local Whisper (CPU/GPU) or Groq hosted.
• AI transcript cleanup.
• Synchronized personal glossary.
• Unified monorepo (desktop, web, backend).

Maintenance & Community

Uses Turborepo for monorepo management; CI/CD workflows defined. Community via Discord: https://discord.gg/5jXkDvdVdt.

Licensing & Compatibility

Licensed under AGPLv3. Strong copyleft requires derivative works to be AGPLv3-licensed, potentially restricting integration into closed-source commercial products.

Limitations & Caveats

Setup involves complex toolchain installs (Node.js, Rust, Tauri) and platform dependencies; Windows GPU builds require Vulkan SDK. AGPLv3 license requires careful evaluation for commercial adoption.