Voice Gecko

What does Voice Gecko do?

Voice Gecko's README describes it as "instant voice-to-text dictation for desktop, free and open source". You press a shortcut, speak, and the text lands on your clipboard. Local speech models (Parakeet, Moonshine and Whisper) are named, and the README says transcripts and dictionary words stay in a local SQLite database on the machine.

The README's account model is unusual: there is a free tier that is the whole product, and an optional paid Support tier that "funds development" but does not unlock features. The README says "Free and Support are the same product, no paywalled features".

Key facts

• Licence: MIT, per the LICENSE file.
• Copyright: Social Freak Limited, trading as Voice Gecko.
• Ready-to-install builds: Windows .msi. The README says macOS and Linux builds are "when available".
• Speech models: Parakeet, Moonshine, and Whisper, all downloaded once and run on the device.
• Local storage: SQLite for transcripts and dictionary words.
• Account: optional, used only for the Support checkout and feedback.
• Stack: React 19, Next.js 15, tRPC, Better Auth, Stripe, PostHog, per the README's own list.
• Monorepo: apps/desktop is a Tauri app; apps/web is the Next.js site with auth, billing and API; packages/* cover api, auth, db, ui, email and payment.
• Build from source: Node 22.14 or later, PNPM 9.6 or later, and Rust for the desktop app. pnpm dev:web runs the Next.js site.
• Whisper GPU sidecar binaries are not in git and are built during packaging (see docs/sidecar.md).

What does it replace, and where does it fall short?

Voice Gecko replaces Wispr Flow for someone on Windows who wants a Whisper-based dictation app with the model running locally, and no account required for the free tier. It is a candidate on the open-source alternatives to Wispr Flow page.

Where it falls short: macOS and Linux builds are not shipped, per the README, so users on those platforms have to build from source (Node, PNPM and Rust). The README says the text lands on your clipboard, so you paste it yourself. For macOS or Linux, see Handy or yap.

How does Voice Gecko run?

The recommended install is the Windows .msi from voicegecko.dev/download or the latest GitHub release, per the README, with no Node, Rust or database setup needed. On first launch, speech models download once, and dictation works offline after that.

To build from source, clone the repo, pnpm install, copy .env.example to .env, and run pnpm dev:web for the Next.js site. The desktop app runs from apps/desktop with pnpm dev (Rust required). Whisper GPU sidecar binaries are built during packaging.

There is no hosted version. The optional paid Support tier funds development and does not unlock features, per the README.

Who is Voice Gecko for?

A Windows user who wants a dictation app with local Whisper, Parakeet or Moonshine speech, no account for the free version, and the option to support the project financially. A user on macOS or Linux who does not want to build from source should look at Handy or yap.

What limits does the README state?

Windows is the only platform with a ready-to-install release build, per the README's line "Install the Windows .msi (macOS/Linux builds when available)". Dictation goes to the clipboard rather than typing at the cursor. The Whisper GPU sidecar binaries are not in git and have to be built during packaging. The account is optional but exists, and paid Support goes through Stripe.

Questions people ask

Is Voice Gecko open source?

Yes. The LICENSE file is MIT, an OSI approved licence, with no additional restriction. The copyright is held by Social Freak Limited, trading as Voice Gecko. Anyone can read, modify and redistribute the code under MIT terms, including for commercial use.

Does Voice Gecko run on Mac or Linux?

Not with a ready-to-install release build. The README says "Install the Windows .msi (macOS/Linux builds when available)", and covers building from source with Node, PNPM and Rust. For a shipping cross-platform alternative that also runs Whisper locally, see Handy or yap.

Does Voice Gecko send my voice to the cloud?

No. The README says dictation runs on-device and lists Parakeet, Moonshine and Whisper as the local speech models. Transcripts and dictionary words stay in a local SQLite database on the machine. The optional account exists only for the Support checkout and feedback.

What does the optional Support tier do?

It funds development, per the README. The README also says explicitly that "Free and Support are the same product, no paywalled features", so the paid tier does not unlock anything. Someone who wants to help the project keep going can pay it; someone who just wants the product free forever can install the Windows .msi and never sign up.

How does Voice Gecko compare with Wispr Flow?

Both turn speech into text on the desktop. Voice Gecko runs Whisper, Parakeet or Moonshine on your own machine, is MIT open source, and ships a Windows installer today. It puts the text on your clipboard for you to paste. Its source code is open, so you can read exactly what it does with your audio. The open-source alternatives to Wispr Flow page has other options.

Sources

• Voice Gecko README and LICENSE: github.com/Lukem121/voicegecko, read 2026-09-28. The README carries the Windows-first release model, the local speech models, and the optional Support tier.