Open Speech Studio

Open source speech recognition for everyone — a local AI alternative to Wispr Flow.

Part of the OpenAEC Foundation.

Features

• 100% Local — All speech recognition runs on your own computer. No cloud, no subscription, full privacy.
• GPU & CPU — Works on both GPU (CUDA/Vulkan) and CPU. Choose the model that fits your hardware.
• Multilingual — Supports 99+ languages with automatic language detection.
• Dictionary — Add your own words, names and jargon for better recognition.
• Universal speech — Works in any application via a global hotkey.
• Open source — Apache 2.0 license. Free to use, modify and distribute.

Installation

Download the latest installer from the Releases page.

Usage

1. Start Open Speech Studio
2. Go to Models and download a speech model
3. Press Ctrl+Shift+Space to start speaking
4. Speak your text
5. Press again to stop — text is automatically inserted

Dictionary

Go to the Dictionary tab to:

• Add your own words (names, abbreviations, technical terms)
• Set up replacements (e.g. "OpenAEC" -> "OpenAEC Foundation")

Models

Via the Models tab you can download models for speech recognition:

Model	Size	Speed	Accuracy	Recommended for
tiny	75 MB	Very fast	Basic	CPU, quick notes
base	142 MB	Fast	Good	CPU, daily use
small	466 MB	Medium	Very good	GPU or fast CPU
medium	1.5 GB	Slow	Excellent	GPU
large-v3	3.1 GB	Very slow	Best	Powerful GPU
large-v3-turbo	1.6 GB	Medium	Near best	GPU

Tech Stack

Component	Technology
Framework	Tauri 2 (Rust)
Frontend	SolidJS + Vite
AI Model	whisper.cpp via whisper-rs
Audio	cpal (Cross-Platform Audio Library)
UI Style	Office-style ribbon (OpenAEC Foundation)

Development

Prerequisites

• Node.js (LTS)
• Rust
• Platform-specific dependencies: see Tauri prerequisites

Getting started

git clone https://github.com/OpenAEC-Foundation/open-speech-studio.git
cd open-speech-studio
npm install
cargo tauri dev

Production build

cargo tauri build

# With GPU support
cargo tauri build --features cuda

Architecture

open-speech-studio/
├── src/                    # SolidJS frontend
│   ├── components/         # UI components (Ribbon, Settings, etc.)
│   ├── lib/                # API bindings
│   └── styles/             # CSS
├── src-tauri/              # Rust backend
│   └── src/
│       ├── lib.rs          # Tauri commands & app state
│       ├── transcriber.rs  # Whisper speech recognition
│       ├── audio.rs        # Microphone recording
│       ├── settings.rs     # Configuration & model auto-detection
│       └── dictionary.rs   # Dictionary
├── models/                 # Whisper models (downloaded via app)
└── index.html              # Entry point

License

Apache License 2.0 — see LICENSE

Contributing

Contributions are welcome! Open an issue or pull request on GitHub.