Download VocalCode
Every install below delivers the same app: push-to-talk voice input for AI coding, with speech recognition that runs on your own machine. The 30-day trial has every feature and asks for no card.
Only vocalcode.app, the winget package DamingWu.VocalCode, the Scoop bucket in wudaming00/vocalcode-docs, and the Homebrew tap wudaming00/vocalcode are official distribution channels.
Windows
Installer
Windows 10 or 11, 64-bit. The installer is Authenticode-signed, and one-click self-update verifies both the manifest hash and the signature before applying an update.
winget
VocalCode is in the Windows Package Manager community repository.
winget install DamingWu.VocalCodeScoop
While the app is too new for Scoop Extras, the manifest lives in an official bucket in the documentation repository.
scoop bucket add vocalcode https://github.com/wudaming00/vocalcode-docs
scoop install vocalcodemacOS
Disk image
macOS 11 or later on Apple Silicon (arm64). The app is Developer ID-signed and notarised by Apple for Gatekeeper.
Homebrew
The official cask lives in the wudaming00/vocalcode tap.
brew install --cask wudaming00/vocalcode/vocalcodeAfter installing
First run
- Hold the talk key or mouse button, speak, release — text lands in the active field in most desktop apps
- The recognition model for your language downloads on demand; after that, recognition works with the wifi off
- macOS asks for Microphone, Accessibility and Input Monitoring, and explains why it needs each one
Licensing
- 30-day trial first, every feature, no card
- $4.99 once at the current launch price, plus applicable tax — no subscription
- One licence covers up to 3 devices
- Already bought? Recover your key
Which languages can it transcribe?
Mandarin Chinese uses a dedicated bilingual Chinese–English Paraformer model. A separate auto-detecting Parakeet model covers 25 European language preferences, currently offered as model-supported previews. See the homepage FAQ for the full list, or the languages and models documentation.