Voxa Dictation Review 2026: Bilingual Voice on Linux

A practical look at Voxa's Omarchy setup, bilingual controls, privacy tradeoffs, and limits.

This Voxa dictation review starts with the project's September 27 Hacker News launch and its focused promise: bilingual push-to-talk dictation for Omarchy, plus an experimental macOS build. That pitch is more useful than another generic "AI voice" app. It tells you where the app runs, how you use it, and which language problem it aims to solve.

The timing makes sense. Linux has plenty of speech-to-text experiments, but far fewer tools that feel like part of the desktop. Multilingual users face another problem. A sentence might start in English, switch to Thai, include a product name, and end with a terminal command. Dictation that works well with clean, single-language audio can break down quickly.

What Voxa dictation is

Voxa is an open-source dictation project built around a native C daemon and command-line tool. It mainly targets Omarchy, a Linux setup based on Arch and Hyprland. The project also includes an experimental macOS path.

On Omarchy, F10 records only while you hold it, and F11 starts or stops recording. When Voxa finishes transcribing, it types the text into the focused app with wtype instead of replacing the clipboard. That works well in terminals, chat boxes, browser forms, and editors, where pasting can sometimes be unreliable.

The app sends audio to ElevenLabs Scribe using your own API key. Voxa supports one primary language and additional secondary languages, with automatic language detection available too. You can also use the keyterms setting to guide transcription toward technical terms, names, and other words the model might otherwise miss.

Why the bilingual setup is interesting

Most multilingual dictation guides tell you to change the system language before you speak. That works when a whole paragraph uses one language, but it gets awkward when people naturally mix languages in the same thought.

Voxa takes a practical approach. You can choose a primary language, add secondary languages, or let it detect the language automatically. Code switching won't always be perfect, but these settings give the transcription service more context than using one fixed language.

Developers and other technical users will get the most from this project. Its README uses Thai and English in the same insertion test, supports custom keyterms, and suits a keyboard-driven desktop. The tool also includes commands for microphone checks, transcription tests, status, and insertion tests. These options make failures easier to diagnose than in an app that simply stops producing text.

What Voxa gets right

The push-to-talk model is the best part. You choose exactly when recording starts and stops, which helps prevent accidental capture and makes short dictation easier to trust.

Direct typing on Omarchy is another smart choice. It leaves the clipboard alone, so anything already copied, such as code, a password manager value, or text for another app, stays there. The project logs state, errors, and timing instead of transcript text.

Setup checks are unusually clear for a young open-source project. voxa doctor, voxa test-mic, and voxa test-scribe help separate microphone, dependency, and transcription problems. If the app doesn't insert text, test the audio path before blaming the application you're using.

The limits matter more than the demo

Voxa isn't an offline dictation tool. It streams audio to ElevenLabs, so you need an internet connection and a paid or trial API allowance. Privacy also depends partly on the retention policy linked to your ElevenLabs account.

The current Omarchy installation also requires Hyprland, PipeWire, wtype, a C compiler, json-c, and a WebSocket-capable libcurl. That makes sense for its target audience, but it isn't a one-click Linux app for everyone.

Recordings stop after 60 seconds. Newlines may work like pressing Enter, so they can accidentally send a chat message or run a terminal command. The project recommends testing in a disposable editor first. Take that warning seriously.

The macOS version is experimental, supports only the default microphone, and pastes text with best-effort clipboard restoration. The author also says they didn't build or test the new macOS path live on the Linux development machine. For now, judge Voxa mainly as an Omarchy tool.

Voxa vs DictaFlow

Voxa makes sense if you use Omarchy, prefer a transparent open source setup, and don't mind supplying your own transcription API key. Its bilingual controls and direct typing on Linux are the main reasons to try it.

DictaFlow fits into more of your daily work. It runs natively on Windows, Mac, iOS, and Android. Hold the talk button to dictate across apps, and app-aware formatting adjusts the cleanup for each one. Custom vocabulary and the Knowledge Base handle names, technical terms, and phrases you use often.

DictaFlow can process speech locally for offline use. You can also choose cloud models when you need better accuracy and formatting. On Windows, it can type into Citrix, RDP, VMware Horizon, EHR fields, and other apps that block clipboard pasting. Pro costs $7/month or $69/year.

The tradeoff is simple. This Voxa dictation review finds a focused, open setup for technical Linux users who want direct API control. DictaFlow is a finished cross-platform app with offline support, app-aware cleanup, and tools for difficult remote desktop workflows. The broader dictation software comparison explains these differences in more detail.

How to test Voxa without risking real work

Install it only after confirming that your libcurl build supports WebSockets. Run voxa doctor, then voxa test-mic. Before testing insertion, run voxa test-scribe to verify the API path.

For the first proper test, open a blank text editor you can discard afterward. Dictate three short samples: one in your primary language, one in your secondary language, and one that switches between the two. Add a product name or technical term to the keyterms list, then dictate the mixed-language sample again.

Then test the places you'll actually use it: a browser form, code editor, chat box, and terminal. Don't test newlines in a live terminal command until you know exactly what Voxa inserts. Measure how long corrections take, not just transcription speed.

If you'd rather use a guided desktop setup than compile and configure a Linux tool, the DictaFlow getting started guide walks you through choosing a hotkey, microphone, model, and text output.

Verdict

The Voxa dictation review comes down to focus. Voxa doesn't try to be a meeting bot, note database, or general voice assistant. You press a key to record, send the audio for transcription, and type the result into the app you're using.

If you use Omarchy, switch between languages, and don't mind managing an API key, it's worth testing. Everyone else should wait for wider Linux packaging and a proven macOS release. Another option is a cross-platform app that already supports desktop, mobile, offline work, and remote sessions.