Local Voice Input

Voice Notes for Mac.
100% Local, 100% Private.

Hold the hotkey, speak, release — your words become a sticky note. Powered by OpenAI Whisper running on the Apple Neural Engine. Audio never leaves your Mac.

macOS 14+  •  Apple Silicon & Intel  •  Voice unlocks with Pro

SlashNote voice-to-note flow on Mac — recorder captures audio, Whisper transcribes locally, and the transcript becomes a structured Client Call Notes sticky with deadlines and action items

How Voice Notes Work on Mac

Three steps. One fluid motion. No app to open, no recording dialog, no upload.

01

Hold the Hotkey

Press and hold your customizable Voice Note hotkey from anywhere on macOS — even when SlashNote is in the background. On first use, SlashNote downloads the right Whisper model for your Mac (~80MB to ~1.5GB depending on hardware).

02

Speak Naturally

Audio is captured at 16 kHz mono PCM. There is no recording length cap for normal use. Speak in any of the supported languages, or let auto-detect pick one.

03

Release. Note Appears.

Whisper transcribes the audio on-device using the Apple Neural Engine. The audio file is deleted. A new sticky note opens with your transcript, ready to edit, color, or pin.

On-Device AI

Powered by Whisper on the Apple Neural Engine

SlashNote runs OpenAI's open-source Whisper speech model fully on-device. Whisper is the same model behind tools like MacWhisper and Superwhisper — but here it runs locally on the Apple Neural Engine, the same silicon Apple Intelligence uses.

On first use, SlashNote picks the best Whisper variant for your Mac and downloads it once. After that, voice notes work fully offline. There is no API key, no quota, no per-minute cost.

  • Open-source Whisper model, MIT-licensed
  • Apple Neural Engine acceleration on Apple Silicon
  • Model auto-selected for your hardware
  • No per-request cost, no rate limits
Whisper model variants
tiny     ~80 MB     fastest, best for clear speech
base     ~150 MB    fast, daily use
small    ~250 MB    balanced speed + accuracy
medium   ~750 MB    high accuracy
large    ~1.5 GB    most accurate, accents & jargon

SlashNote auto-selects the right variant for your Mac. Downloaded once, cached locally, works offline thereafter.

Two Ways to Capture a Voice Note

Pick the entry point that matches what you are doing.

⌨️

Global Voice Note Hotkey

Push-to-talk from anywhere on macOS. Hold your customizable hotkey, speak, release — a new sticky note opens with your transcript. Works even when SlashNote is hidden in the background.

Customize the hotkey in Settings → Shortcuts → Voice Note. Pro feature.

🎙️

In-Note Voice Button

Already in an existing note? Tap the microphone in the editor footer to dictate more content into the same note. Same engine, same on-device path — just appended to what you have.

Available inside the SlashNote editor. Pro feature.

Your Audio Never Leaves Your Mac

Local-first by construction, not by policy.

🎚️

Captured Locally

Audio is recorded at 16 kHz, 16-bit, mono PCM. The microphone is only active while you hold the hotkey.

🧠

Transcribed On-Device

Whisper runs on the Apple Neural Engine. No network call, no cloud round-trip, no analytics on what you said.

🗑️

Audio File Deleted

The temporary audio file is removed immediately after the transcript is produced. Only the text remains.

✈️

Works Offline

After the first-use model download, voice notes work without internet. On a plane, in a tunnel, in a faraday cage.

See the full privacy policy or read why privacy-first note-taking matters.

Multilingual Voice-to-Text

The Whisper model supports 100+ languages. SlashNote exposes the 10 most-used in settings, plus auto-detect — which routes to any of the 100+.

English
Russian
German
French
Spanish
Italian
Portuguese
Japanese
Korean
Chinese

Pick a specific language for highest accuracy on uncommon vocabulary, or leave it on auto-detect and switch mid-thought.

Accuracy Tuned to Your Mac

SlashNote auto-selects the best Whisper variant for your hardware. You do not pick a model — your Mac does.

Smaller Macs

Tiny and base variants (~80MB to ~150MB). Fast transcription, good for short captures and clear speech.

⚖️

Mid-Tier Apple Silicon

Small and medium variants (~250MB to ~750MB). Balanced accuracy and speed for everyday voice notes.

🎯

Higher-End Apple Silicon

Large variant (~1.5GB). Top-tier accuracy on accents, technical vocabulary, and long-form speech.

Apple Silicon (M1 and later) recommended — Whisper accelerates on the Apple Neural Engine. Intel Macs run on CPU and are supported on macOS 14+.

Voice Notes vs Siri Dictation vs Cloud Dictation

Same input — three very different paths your audio can take.

Aspect SlashNote Voice Notes Apple Dictation Cloud Dictation Apps
Processing 100% on-device (Whisper) On-device (Apple Silicon) Cloud servers
Internet Required No (after model download) Sometimes Always
Output Structured sticky note Text into focused field Text into app of choice
Languages (UI selectable) 10 + auto-detect (model: 100+) ~40 system languages Varies by app
Audio Retention Deleted after transcription Not stored Provider-dependent
Cost Pro plan, no per-minute fee Free with macOS Subscription or per-minute
Best For Quick voice → searchable notes Dictating into apps Long meetings, transcripts

Deeper dive in our blog: on-device vs Siri vs cloud voice — a technical comparison.

Voice Into a Note, Not Just a Text Field

Dedicated dictation apps are great for typing-by-voice. SlashNote is built for capturing thoughts that need to live somewhere.

A Dictation App

Inserts raw transcribed text into whichever field your cursor is in. The output is ephemeral — once you close the document, you have to remember where you put the thought.

  • Output: raw text in a focused field
  • No structure, no organization
  • No search across captures
  • No AI on top of what you said

SlashNote

Creates a structured sticky note you can pin over windows, color-code, search, and run AI on. Your MCP server can read it. Claude or Cursor can summarize or restructure it. The thought has a home.

  • Output: a real note, persistent and searchable
  • Pin over windows, color, organize
  • Searchable across every voice capture
  • AI modifiers and MCP-accessible

Want the full voice-to-AI workflow? Read the AI voice notes workflow guide.

What People Use Voice Notes For

Different contexts, same gesture: hold, speak, release.

🎙️

Meeting Recaps

Speak the highlights right after a meeting — names, decisions, next steps. Saved as a sticky you can pin while you write the follow-up.

🚶

Walking Brainstorms

Capture ideas on a walk without taking your hands off your coffee. Hold the hotkey, talk, release — the thought is a note.

📓

Journaling

End-of-day reflections in your own voice, transcribed locally. Audio never leaves the Mac; the text stays in your notes.

💻

Code TODOs

Explain a refactor idea out loud while it is fresh, without leaving your editor. The note lands ready to act on next session.

🌍

Language Practice

Switch input language in settings or use auto-detect. Speak in any of 100+ languages Whisper supports.

Accessibility

Voice-first capture for anyone who finds typing tiring or impractical — fully on-device, no account required.

Requirements

System

  • macOS 14 (Sonoma) or later
  • Apple Silicon recommended (M1+)
  • Intel Macs supported (CPU transcription)
  • ~80MB to ~1.5GB disk for the Whisper model

Plan

  • Voice input unlocks with Pro
  • Pro Monthly: $9.99/mo
  • Pro Yearly: $29.99/yr
  • Lifetime: $29.99 one-time
See Pro plans →

Frequently Asked Questions

Is voice input on Mac private with SlashNote?
Yes. SlashNote runs OpenAI Whisper entirely on the Apple Neural Engine. Audio is captured locally, transcribed on your Mac, and the audio file is deleted immediately after transcription. No audio is ever sent to any server — not to SlashNote, not to OpenAI, not to anyone.
Does voice input work offline?
Yes, once the Whisper model is downloaded on first use. After that, voice notes work with no internet connection at all. The model is fetched once and cached locally; subsequent uses are fully offline.
Is voice input free or a Pro feature?
Voice input is a Pro feature, included in Pro Monthly ($9.99/mo), Pro Yearly ($29.99/yr), and Lifetime ($29.99). The free tier of SlashNote includes unlimited notes, AI requests, and the global hotkey for text notes.
What languages does voice input support?
The underlying Whisper model supports 100+ languages. SlashNote settings expose 10 of the most-used (English, Russian, German, French, Spanish, Italian, Portuguese, Japanese, Korean, Chinese) plus auto-detect. Auto-detect routes any language the underlying model recognizes.
Do I need Apple Silicon for voice notes?
Apple Silicon (M1 and later) is recommended for the fastest transcription, since Whisper runs on the Apple Neural Engine for acceleration. Intel Macs running macOS 14 are supported but transcribe on the CPU, which is slower.
How accurate is the transcription?
Accuracy depends on the Whisper model variant your Mac runs (tiny, base, small, medium, or large — auto-selected based on hardware). On Apple Silicon, accuracy is comparable to leading cloud speech services for most languages. Background noise, accents, and technical vocabulary affect results as with any speech-to-text engine.
What's the difference between SlashNote and apps like MacWhisper?
MacWhisper and similar dedicated dictation apps focus on transcribing audio files or system-wide dictation into any text field. SlashNote is a notes app first — voice produces a structured sticky note you can pin, color-code, search, AI-process, and access from Claude or Cursor via the MCP server. Different product category, similar underlying engine.
Where is the voice model stored on my Mac?
Voice models are cached locally in SlashNote's application support folder. Depending on your hardware, the Whisper model is ~80MB (tiny) to ~1.5GB (large), downloaded once on first use of voice input.

Try Voice Notes for Free

Download SlashNote — free on the Mac App Store. Voice input unlocks with Pro.

Download on the App Store