Voice to Text,
Fully Offline

WHISPER \ LOCAL LLM \ PRIVACY \ macOS

Free, open-source AI voice dictation for macOS. Press a hotkey, speak naturally, and get polished text pasted into any app. No cloud, no subscription.

How it works

1

Press hotkey

Alt + Space

2

Speak naturally

Audio stays local

3

Whisper transcribes

On-device STT

4

LLM refines & pastes

Into any app

Key features

100% offline & private

Everything runs locally on Apple Silicon. Zero network requests. Your voice never leaves your machine.

No subscription

Free forever. No account, no API keys, no usage limits, no word caps.

Whisper + local LLM

Transcription via Whisper (base model, 142 MB). Refinement via Qwen 1.5B or 4B, running through llama.cpp.

Auto-paste anywhere

Polished text is pasted directly into the focused app. Works in any text field across macOS.

Swift + Rust

SwiftUI menu bar app with a Rust core for Whisper and LLM inference. Built for Apple Silicon performance.

Open source (MIT)

Fully open source under the MIT license. Inspect, modify, and contribute on GitHub.

Comparison

PhemyWispr FlowWillowAqua Voice
PriceFree$12/mo$12/mo$8/mo
Open sourceYesNoNoNo
Fully offlineYesNoNoNo
Voice to LLMYesYesYesYes
Auto-pasteYesYesYesYes
No account neededYesNoNoNo
Data stays localYesNoNoNo

Requirements

  • macOS 14+ (Sonoma or later)
  • Apple Silicon (M1 or later)
  • ~1–3 GB disk space for models (downloaded on first launch)

Ready to try Phemy?

Download the latest release or build from source.