Voice Gecko: A Free, Open-Source Dictation App Built for Windows
Voice Gecko is a free, open-source, on-device dictation app for Windows: hold a shortcut, speak, and the text lands on your clipboard, with no subscription and no Mac requirement.
Most of the on-device dictation apps getting attention lately are Mac-only. Phantom Voice, covered here recently, only runs on Apple Silicon and charges a one-time $10. Voice Gecko does the same core job on the platform Phantom Voice doesn't touch: it's a free, open-source dictation app for Windows, built by developer Luke M, and it doesn't gate a single feature behind payment. Hold a shortcut, speak, and the transcribed text lands on the clipboard, ready to paste into an email, a code editor, or an AI chat window.
What Voice Gecko actually does
The mechanic is familiar to anyone who's used a push-to-talk dictation tool: press a keyboard shortcut, speak, release, and the text shows up on the clipboard. Voice Gecko runs this through a persistent interface called the GeckoBar, which stays available across whatever app currently has focus, whether that's Slack, VS Code, Jira, Discord, Telegram, or GitHub. Everything happens on-device. No audio leaves the machine, and there's no cloud account required to use it, which is the same privacy pitch every serious local dictation app makes, but here it's paired with a fully inspectable, MIT-licensed codebase on GitHub rather than a closed binary asking to be trusted.
Transcripts are stored locally in SQLite, so a history of what's been dictated lives on the machine, not on a server somewhere. The product also ships a small local dictionary feature, letting users teach it names, acronyms, or jargon that a general-purpose speech model tends to mangle on the first pass.
Three transcription engines, not one
Where Phantom Voice standardized on whisper.cpp with a choice of four Whisper model sizes, Voice Gecko takes a different approach: it supports multiple local speech models, including Parakeet, Moonshine, and Whisper, and lets the underlying engine vary rather than locking into a single family. Parakeet and Moonshine are both newer, lighter-weight local speech models built for exactly this kind of low-latency, on-device transcription task, distinct from Whisper's architecture. Having more than one engine option is useful in practice for the same reason model choice matters anywhere else: a smaller model trades some accuracy for speed, and different engines handle accents, background noise, or technical vocabulary differently. Speech models download once during initial setup and then run entirely offline afterward.
Genuinely free, not free with a catch
This is where the real gap against Phantom Voice opens up. Phantom Voice is a $10 one-time purchase with a 14-day trial. Voice Gecko is free, full stop, and its own documentation is explicit about what that means: there's an optional $5.99-a-month "Support" tier for people who want to fund ongoing development, but the project states plainly that free and paid are the same product, no paywalled features either way. That's a meaningfully different posture from most of the dictation app category, including tools that call themselves free while holding back a model size, a word limit, or a cleanup feature behind a paywall. Voice Gecko's entire feature set, all three transcription engines, the local dictionary, the clipboard workflow, is available without paying anything.
The source is MIT-licensed on GitHub with no carve-out for branding or logos, which is a slightly more permissive license than Phantom Voice's MIT-minus-branding approach. For a solo open-source project, that's a real commitment: anyone can fork the whole thing, name and all, and build on it.
Windows only for now, with Mac and Linux planned
The platform story is the other clear differentiator, and it cuts the opposite way from Phantom Voice. Voice Gecko currently ships as a Windows installer only. Phantom Voice, by contrast, requires macOS 14 or newer on Apple Silicon specifically, with no path to Intel Macs, Windows, or Linux. Voice Gecko's own GitHub README lists "macOS/Linux builds when available" as a stated direction, not a shipped reality, so anyone on a Mac still needs to look elsewhere today, likely at Phantom Voice or a paid tool like WhisperFlow itself. But for the much larger population of Windows users who've watched this entire category of on-device, Whisper-based dictation app ship for Mac first and Windows never, Voice Gecko is one of the first credible, actively maintained open-source entries built for their platform from day one.
Who it's for, and who it's not
Voice Gecko is aimed at the same core user as any push-to-talk dictation tool: people who think faster than they type, especially through long prompts, emails, or first drafts, and who'd rather not hand a continuous audio stream to a cloud service to get that speed. The multiple-engine setup gives it a slight edge for anyone who wants to experiment with model tradeoffs instead of accepting whatever one engine a closed app picked for them.
It's explicitly not for Mac users today, and it's not a mature project yet either. The GitHub repository sits at a modest star count, which is normal for a tool a few months past its Product Hunt launch, but it does mean fewer eyes have stress-tested it against Phantom Voice's more established (if still small) user base. Anyone evaluating it should expect the rough edges of an early open-source project: less polish, a smaller community for troubleshooting, and features that may shift as development continues.
Current traction
As of today, Voice Gecko sits at 104 upvotes on Product Hunt, where it launched with a #4 daily ranking. That's a modest number, and notably lower than some earlier estimates floating around, worth checking the live page directly rather than trusting a cached figure. The traction is smaller than Phantom Voice's, but the two aren't really competing for the same install: one requires a Mac and a one-time payment, the other requires Windows and nothing at all. For anyone dictating on Windows who wants the on-device privacy story without opening their wallet, Voice Gecko is a genuine, if early, option.
Source: Voice Gecko on Product Hunt, voicegecko.dev, github.com/Lukem121/voicegecko
Join the newsletter
AI workflows and systems, straight to your inbox.
No spam. Unsubscribe anytime.