Licenses
Last updated: 16 October 2026
Hear is free to use. It stands on the work of many open projects, and this page thanks them and lists the license each one comes under. Hear itself is © Liam, all rights reserved.
AI models
The models are downloaded when you install or turn them on. They are not part of Hear’s program. Several were converted to a format Hear can run (ONNX or GGUF) and compressed (quantized) by the projects named under “Converted by”. The weights are otherwise unchanged.
| Model | Used for | License | Original · Converted by |
|---|---|---|---|
| Qwen3-ASR 1.7B | Ultra dictation | Apache-2.0 | Qwen · ggml-org |
| Qwen3.5 4B | The assistant | Apache-2.0 | Qwen · Unsloth |
| Parakeet TDT 0.6B v3 | Fast dictation | CC BY 4.0 | NVIDIA · istupakov |
| Canary 1B v2 | Precise dictation | CC BY 4.0 | NVIDIA · istupakov |
| Whisper tiny | Finding the spoken language | Apache-2.0 | OpenAI · onnx-community |
| Silero VAD | Detecting speech and silence | MIT | Silero · istupakov |
Parakeet TDT 0.6B v3 and Canary 1B v2 are © NVIDIA, licensed under CC BY 4.0. They were converted to ONNX and quantized by the istupakov project. Hear is not endorsed by NVIDIA.
AI engine
Libraries
Hear is written in Python. These are the libraries it uses directly. Each one also depends on other open projects; the full list is available on request.
| Library | Used for | License |
|---|---|---|
| Python | The language Hear runs on (including Tk, for the on-screen pill) | PSF-2.0 |
| onnx-asr | Runs the Fast and Precise models | MIT |
| ONNX Runtime | Runs ONNX models | MIT |
| NumPy | Audio calculations | BSD-3-Clause |
| SciPy | Optional noise reduction | BSD-3-Clause |
| sounddevice | Microphone access | MIT |
| keyboard | Shortcut keys and typing | MIT |
| Pyperclip | Clipboard | BSD |
| Pillow | Drawing the pill and the cards | MIT-CMU |
| pystray | The icon next to the clock | LGPL-3.0 |
| uiautomation | Finding the text field and selected text | Apache-2.0 |
| pywebview | The settings window | BSD-3-Clause |
| ddgs | Web search for the assistant | MIT |
pystray is used as a separate library, installed on its own next to Hear, so you can replace it with another version of your choice as the LGPL allows.
Services
- Weather data by Open-Meteo.com, under CC BY 4.0. Used for non-commercial purposes: Hear is free, with no ads and no subscription.
- Web search by DuckDuckGo. Models are hosted by Hugging Face; the engine and updates by GitHub.
Trademarks
Microsoft, Windows and the Microsoft Store are trademarks of the Microsoft group of companies. Qwen, NVIDIA, Parakeet, Canary, Whisper and the other names on this page belong to their respective owners. They are mentioned only to say which projects Hear is built with. Hear is not affiliated with, sponsored by or endorsed by any of them.
A mistake or a missing credit? Please tell us at github.com/LiamLitle/hear-site, and we will fix it.