| Price and license | $50 once after the 14-day full trial; commercial proprietary app | Free and MIT licensed, with public source code |
|---|
| Mac requirements | macOS 26+ on an Apple-silicon Mac | macOS 26+ on Apple silicon |
|---|
| Speech and cleanup | Local speech recognition plus supported on-device Apple formatting in two fixed modes | Streaming Apple SpeechAnalyzer transcription plus on-device Foundation Models cleanup, with deterministic Basic cleanup as fallback |
|---|
| Languages | Nine English regional speech choices | Multiple Apple on-device speech languages; the current product site lists 30 locale choices |
|---|
| Personalization and context | Fixed Dictate and Voice-to-Spec modes; optional local repository context only inside Voice-to-Spec | Personal Dictionary, automatic learning, JSON import or export, recent-text context, and active app or window-title context processed locally |
|---|
| Workflow | Dictate plus Voice-to-Spec Markdown that can be reviewed before guarded Codex or Claude Code handoff | General cursor dictation, hold or toggle activation, voice macros, and “Hey Megaphone” generation or rewriting |
|---|
| Saved audio and video | Does not transcribe imported files | The reviewed first-party product workflow documents live dictation, not a saved-file transcription feature |
|---|
| Installation trust | Developer ID signed, Apple-notarized DMG with Gatekeeper verification, plus a Homebrew cask | Signed DMG, but the publisher says it is not notarized and may require clearing quarantine or choosing Open Anyway; source builds are available |
|---|
| Accounts and network | No account for the trial; speech stays local; paid use performs a Gumroad license check | No account, subscription, transcription API, API key, or Megaphone server; first-use Apple speech-model download still applies |
|---|