Mac has come a long way with voice-to-text. Between Apple's built-in dictation and third-party apps using OpenAI's Whisper, you can now get highly accurate transcription without ever sending your voice to the cloud.
This guide covers everything you need to know to set up and optimize voice-to-text on your Mac.
Option 1: Apple's Built-in Dictation
macOS Sonoma and later include on-device dictation. To enable it:
- Open System Settings → Keyboard → Dictation
- Toggle Dictation on
- Choose your language and shortcut key
Pros: Free, built-in, on-device processing in Sonoma+
Cons: No personal dictionary, no grammar correction, limited customization, no file transcription
Option 2: Whisper-Based Apps
OpenAI's Whisper is an open-source speech recognition model that runs locally on Apple Silicon. Several Mac apps use it:
Push-to-Talk Dictation (VoxBee, SuperWhisper, Sotto)
These apps let you hold a hotkey, speak, and release — your words appear wherever your cursor is. VoxBee adds grammar correction, filler word removal, and a personal dictionary. It also includes file transcription and meeting recording.
File Transcription (VoxBee, MacWhisper)
Drop an audio/video file or paste a URL to get a full transcript. VoxBee supports 1,800+ sites including YouTube, podcasts, and social media.
Choosing the Right Whisper Model
Whisper comes in multiple sizes. Bigger models are more accurate but slower:
- Tiny (75MB) — Fastest, good for quick notes in quiet environments
- Base (140MB) — Good balance for everyday dictation
- Small (460MB) — Noticeably better accuracy
- Medium (1.5GB) — Great accuracy, still responsive on M1+
- Large v3 (2.9GB) — Best accuracy, best for recordings and meetings
For live dictation, start with Base or Small. For transcribing files and meetings, use Large v3.
Tips for Better Accuracy
- Use a good microphone — Your Mac's built-in mic works, but a dedicated mic (even AirPods) improves accuracy significantly
- Minimize background noise — Close windows, use noise-canceling headsets for meetings
- Speak naturally — Whisper handles natural speech better than slow, deliberate dictation
- Use a personal dictionary — Add names, technical terms, and domain-specific words your app keeps getting wrong
- Pick the right model — Larger models handle accents, technical terms, and noisy environments better
Privacy Considerations
The biggest advantage of local voice-to-text is privacy. With apps like VoxBee, your audio never leaves your Mac. This matters for:
- Confidential business conversations
- Medical or legal dictation
- Personal journals and notes
- Working in regulated industries (HIPAA, GDPR)
Cloud-based services like Wispr Flow and Otter.ai process your audio on remote servers, which means your voice data passes through third-party infrastructure.
Getting Started
The fastest way to get set up with local voice-to-text on Mac:
- Download VoxBee (14-day free trial, no account needed)
- Pick a Whisper model (start with Base for speed or Large v3 for accuracy)
- Hold Option, speak, release — text appears at your cursor
That's it. No account, no internet, no configuration needed.