rccyx

rccyx / asryx

Public

Native C++ ASR for Linux

16
1
100% credibility
Found May 31, 2026 at 16 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
C++
AI Summary

asryx is a voice-to-text tool for Linux computers that works completely offline. You press a button to start recording your voice, press it again when you're done, and the program converts your speech into written text that appears in your clipboard. It uses a local language model called whisper.cpp, so nothing is ever sent to the internet. The program handles all the technical details automatically: it picks the right audio recorder for your system, manages temporary files, sends you notifications when recording starts and finishes, and cleans up after itself. You can choose from several language models depending on how accurate you need the transcription to be, and you can configure which language you're speaking to speed things up.

How It Works

1
🎤 You want to turn speech into text

You discover asryx and learn it can transcribe your voice into written words, right on your own computer, without sending anything to the internet.

2
📦 You install it once

A simple installer sets everything up automatically, downloads a small language model, and places the program where your computer can find it.

3
⌨️ You assign a keyboard shortcut

You pick a key combination you like (like Alt+W) and bind it to the program in your desktop settings.

4
🔴 You press the shortcut to start recording

A small notification appears telling you it's listening. You can speak naturally while it captures everything.

5
🔇 You press the shortcut again to stop

The program finishes listening, processes your words through the language model, and prepares your transcript.

6
Your transcript appears
💬
Paste into a document

You press Ctrl+V in your notes, email, or message and your spoken words appear as text.

📱
Paste into another app

You switch to any app and paste your transcript wherever you need it.

✨ Your words are now text

Everything was processed privately on your machine, the temporary files cleaned up, and you're ready for your next recording.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 16 to 16 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is asryx?

asryx is a native C++ speech-to-text tool for Linux that runs entirely offline. You press a keybind once to start recording, press it again to stop, and the transcript lands in your clipboard. No server calls, no API keys, no Python runtime, no containers. It uses whisper.cpp as the recognition engine, captures audio through PipeWire or ALSA depending on your setup, writes to wl-copy or xclip for clipboard output, and sends notifications when it's done. Model management happens through a simple CLI: list available models, install sizes from tiny to large, switch between them, and set language preferences.

Why is it gaining traction?

The offline-first approach is the hook. Everything runs locally on your machine with no network dependency during transcription. The toggle UX is refreshingly simple - it's literally just a binary that flips between recording and transcribing states. Lock files prevent race conditions from key repeat or compositor double-fire. The status output integrates cleanly into Waybar and Polybar, letting you surface recording state in your panel. For developers who value privacy or work in low-connectivity environments, a self-contained C++ binary with no runtime dependencies is more trustworthy than yet another Python package.

Who should use this?

Linux power users on tiling window managers like Hyprland, Sway, or i3 who want a keyboard-driven voice-to-text workflow. Developers handling sensitive audio who cannot send data to cloud services. Anyone frustrated with Python-based ASR tools and their dependency chains. Not suitable if you need macOS or Windows support - this is Linux-only, and the project has not reached maturity.

Verdict

The design is solid and the offline model is genuinely useful. With a 1.0% credibility score and only 16 stars, the project is very early-stage - docs are minimal and test coverage is unknown. Try it if you run Linux and want local speech-to-text. Do not adopt it for production without evaluating the maintenance trajectory first.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.