Zerm

GPLv3 macOS dictation

Zerm turns speech into clean text on macOS.

Tap a hotkey, speak, and paste. Zerm is a native macOS dictation app based on VoiceInk by Beingpax, adapted for Arcusis branding, product direction, and ongoing GPLv3 development.

latest release GPLv3 licensed VoiceInk credited
01 Record

Use a global macOS shortcut or push-to-talk style recording.

02 Transcribe

Use local and explicitly configured transcription engines.

03 Paste

The result lands in your clipboard, with optional native auto-paste on macOS.

Local first Local transcription paths keep audio on the machine.
VoiceInk lineage Built from the GPLv3 VoiceInk foundation with public credit.
Auditable Swift, SwiftUI, transparent source, and GPLv3 license terms.

Native writing layer

Built around the OS behaviors that make dictation reliable.

Zerm focuses on the macOS behaviors that matter for fast dictation: permissions, active app context, insertion state, model selection, recording feedback, and reliable paste behavior.

01 macOS auto-paste

Uses native macOS permissions and insertion paths to paste the transcript at your cursor.

02 Power Mode context

Detects active apps and browser URLs so prompts can adapt to the place you are writing.

03 Visible permissions

Surfaces microphone, Accessibility, and screen-context requirements instead of hiding setup failures.

Read native verification checklist

Downloads

Install the macOS app or compile from source.

Download links are populated directly from the latest release. Current version: loading….

Origin and license

Based on VoiceInk, credited in public, and licensed under GPLv3.

Zerm is a modified derivative of VoiceInk by Beingpax. VoiceInk provided the foundation for the native macOS dictation workflow, transcription pipeline, Power Mode, model handling, and supporting services. Zerm keeps the GPLv3 license and links back to upstream.

Privacy model

Clear boundaries for local and configured services.

Zerm records through macOS microphone APIs, supports local transcription paths, and only uses cloud or AI providers that the user configures. Context-aware features may require additional macOS permissions so the app can understand where text is going.

  • Local transcription paths keep audio on the device.
  • Cloud transcription providers require explicit setup.
  • AI request payload storage is controlled separately from runtime requests.
  • Power Mode uses active app and browser context only for configured workflows.
  • macOS auto-paste requires Accessibility permission.

Open source

Read the code, file issues, send patches.

The README includes VoiceInk attribution, GPLv3 licensing, architecture notes, build instructions, and contributor guidance.

Open repository