Private dictation into any Mac app.

Hold a key, speak, release. Presspeech pastes text at the cursor in about 100 ms with no account, no subscription, and no cloud transcription. Apple Silicon, macOS 14+.

The current Presspeech identity migrates saved preferences and local dictionary rules from earlier builds. macOS privacy permissions require one fresh grant.

brew install --cask rcourtman/presspeech/presspeech
~100 mskey release to pasted text
2.5 MBsigned release zip
~80 MBRAM while idle
0% CPUbetween dictations
The whole workflow

Hold. Speak. Release.

Animated demo: hold Right Option, speak, release — the sentence appears at the cursor about 100 milliseconds later, with no network round-trip.
New in 0.3

Say exactly what should happen

Voice shortcuts

Map a phrase you say to exact reusable text. No rewriting model, no cloud call, and no ambiguity about the result.

Spoken formatting

Opt in to deterministic commands including “new paragraph”, “bullet point”, “comma”, “open quote”, and more.

Try Dictation

Finish setup in a focused private scratchpad, test the hotkey immediately, then take the same workflow into any Mac app.

One workflow

Built for one thing, not ten

Press and speak

Presspeech is a menu-bar utility, not a document manager. The default trigger is Right Option, with other hotkeys available from the menu.

Local speech recognition

Audio is captured in memory, transcribed locally, pasted, then discarded. Transcripts are not written to disk.

Native Mac stack

Swift, AVFoundation, CoreGraphics, AppKit, CoreML, and FluidAudio. No embedded browser runtime, interpreter, account, subscription, or cloud transcription endpoint.

Good fit

When Presspeech is the right pick

Use Presspeech when you want

Fast phrase-length dictation into whatever app already has focus; a local-only privacy model; a tiny notarised app; and a workflow that does not keep a transcript library.

Use something else when you need

Intel Mac support, macOS before 14 Sonoma, batch file transcription, speaker labels, long-form transcript editing, cloud models, team features, or integrations with notes databases.

Why trust it

Evidence and install paths

BenchmarkedFluidAudio on the Apple Neural Engine is measured against the prior GPU path on the same WAV clips.
AuditableThe app is a single Swift source file plus release scripts, signed GitHub release assets, and a public Homebrew Cask tap.
Private by designThe full network surface is documented: model download, update check, and user-triggered install/update downloads.
Agent-readableInstall instructions, benchmark results, privacy facts, and comparison pages are available as plain HTML, Markdown, JSON, and text.