Vocalinux - Offline Voice Dictation for Linux

TL;DR

TL;DR: Vocalinux is a free, open-source voice dictation tool for Linux that runs 100% offline using whisper.cpp, VOSK, or OpenAI Whisper — no internet required, no data leaves your machine.

Source and Accuracy Notes

⚠️ This section is MANDATORY. All links must be verified from actual source, not guessed.

• Official site: vocalinux.com — visited and verified
• Source repository: github.com/jatinkrmalik/vocalinux — README read in full
• License: GPL-3.0 — verified via repository
• HN launch thread: news.ycombinator.com/item?id=47040707
• Source last checked: 2026-08-03 (commit main branch, README v0.15.0)

What Is Vocalinux?

Linux has always punched above its weight — except when it comes to voice typing. Vocalinux fixes that.

It’s a free, GPLv3-licensed desktop app that lets you dictate text into any application on Linux, using fully offline speech recognition. Pick from three engines (whisper.cpp, OpenAI Whisper, or VOSK), get automatic GPU acceleration via Vulkan, and control it with customizable keyboard shortcuts: toggle or push-to-talk.

The project was launched on Hacker News and has accumulated 707 GitHub stars as of August 2026.

Setup Workflow

Step 1: Interactive Install (Recommended)

The installer auto-detects your hardware (GPU, RAM, Vulkan support) and recommends the best engine:

curl -fsSL raw.githubusercontent.com/jatinkrmalik/vocalinux/main/install.sh -o /tmp/vl.sh && bash /tmp/vl.sh

The installer downloads the default whisper.cpp tiny model (~74MB) and completes in 1–2 minutes.

Step 2: Choose Your Engine

| Engine | Command | GPU Support | Install Time | |--------|---------|-------------|--------------| | whisper.cpp (default) | bash /tmp/vl.sh | AMD/Intel/NVIDIA via Vulkan | ~1–2 min | | OpenAI Whisper | bash /tmp/vl.sh --engine=whisper | NVIDIA only (PyTorch + CUDA) | ~5–10 min | | VOSK | bash /tmp/vl.sh --engine=vosk | CPU only | ~40MB model |

Step 3: Launch and Dictate

After installation, Vocalinux runs as a system tray app. Use the toggle or push-to-talk hotkey to start dictating — text is injected directly into any focused application.

Step 4: Arch Linux (AUR)

yay -S vocalinux

Deeper Analysis

Three Speech Recognition Engines

Vocalinux ships with three engine backends:

• whisper.cpp — Default. C++ implementation with universal GPU support via Vulkan. Works on AMD, Intel, and NVIDIA GPUs. Fast, low-memory footprint.
• OpenAI Whisper — PyTorch-based. Requires NVIDIA GPU with CUDA. Larger model downloads (~1–3GB), longer install time.
• VOSK — Lightweight. Works on systems with as little as 4GB RAM. Useful for older hardware.

GPU Acceleration via Vulkan

Unlike many Linux voice tools that only support NVIDIA CUDA, Vocalinux uses Vulkan compute shaders for GPU acceleration. This means AMD and Intel integrated/discrete GPUs get hardware acceleration without proprietary CUDA dependencies.

Display Server Compatibility

Vocalinux supports both X11 and Wayland compositors. On Wayland, it uses IBus integration for text injection into applications. The installer detects which display server is running and configures accordingly.

Version 0.15.0 Highlights

The latest release (v0.15.0) adds:

• Searchable settings sidebar
• AppImage packages (self-contained x86_64 and aarch64)
• ~33 selectable speech languages
• Auto-capitalization after sentence punctuation
• Vulkan GPU selection (prefer discrete GPUs automatically)
• Idle timeout model unloading for power saving

Practical Evaluation Checklist

• [ ] Interactive installer auto-detects GPU and Vulkan support
• [ ] Microphone input works on first launch
• [ ] Text injection works in at least one X11/Wayland application
• [ ] System tray icon shows listening state
• [ ] Hotkey toggle / push-to-talk functions correctly
• [ ] Settings survive a restart
• [ ] whisper.cpp engine runs without internet

Security Notes

Privacy by design: Vocalinux is 100% offline. No audio data is ever transmitted over the network. This is verified from the README:

“No internet required. No data leaves your machine. Just speak and type.”

The tool uses local speech recognition models stored on-disk. No telemetry, no cloud sync, no API calls to external services.

FAQ

Q: Does it work on headless servers? A: Vocalinux is a desktop GUI application designed for systems with a display server. It has not been tested in headless server environments.

Q: Can I use it without a GPU? A: Yes. All three engines (whisper.cpp, Whisper, VOSK) run on CPU. VOSK is the lightest option for low-RAM systems (4GB).

Q: Does it work with any microphone? A: Vocalinux uses PulseAudio or PipeWire for audio capture on Linux. Most USB and built-in microphones are supported.

Q: How large is the default model? A: The default whisper.cpp tiny model is approximately 74MB. Larger models (small, medium) offer better accuracy at the cost of more RAM and compute.

Q: Does it work alongside other voice tools like Whisal? A: Both can run simultaneously, but only one should capture microphone input at a time to avoid conflicts.

Conclusion

Vocalinux fills a genuine gap in the Linux desktop ecosystem: a privacy-respecting, GPU-accelerated, offline-first voice dictation tool that works on both X11 and Wayland. The one-line installer and automatic hardware detection make it accessible to users who do not want to hand-configure speech recognition pipelines.

For users who prioritize privacy above all else, Vocalinux’s zero-network design is a clear differentiator. For developers, the three-engine architecture (whisper.cpp, Whisper, VOSK) provides flexibility to swap engines without changing the UI.

Try it:

curl -fsSL raw.githubusercontent.com/jatinkrmalik/vocalinux/main/install.sh -o /tmp/vl.sh && bash /tmp/vl.sh