How to Use Speech to Text on Mac: Complete Guide (2026)
How to Use Speech to Text on Mac: Complete Guide (2026)
Every Mac ships with speech to text built in, but Apple's version tops out at about 90% accuracy and cuts you off after 60 seconds. Knowing how to use speech to text on Mac properly — and when to move beyond the built-in option — is the difference between a tool you abandon after a week and one that genuinely replaces typing.
What speech to text on Mac actually does
Speech to text converts your spoken words into typed text in real time. On a Mac, this happens through either Apple's built-in Dictation feature or a third-party app that handles the audio capture, transcription, and text insertion independently.
The technology has shifted dramatically since 2023. Apple Silicon Macs (M1 through M4) run speech recognition models directly on the Neural Engine — a dedicated chip designed for machine learning inference. This means transcription happens on your device, not on a remote server. The result is lower latency, better privacy, and no dependency on your internet connection.
For Mac users in 2026, the practical question isn't whether speech to text works. It does. The question is which tool gives you the accuracy, speed, and integration that makes it worth using every day.
How to enable Apple Dictation on Mac
Apple Dictation is the starting point for most Mac users. It's free, already installed, and works in most apps that accept text input.
Step-by-step setup
- Open System Settings (click the Apple menu in the top-left corner, then System Settings)
- Navigate to Keyboard in the sidebar
- Scroll down to the Dictation section
- Toggle Dictation to On
- Choose your language — this determines which speech model loads, and a wrong choice tanks accuracy silently
- Set your keyboard shortcut (default is double-tap Fn, but you can change it to any modifier key)
- Select your microphone source from the dropdown below the toggle
That's it. Dictation is now active system-wide.
How to start dictating
Place your cursor where you want text to appear — an email draft, a Notes document, a text field in any app. Press your dictation shortcut (double-tap Fn by default). A small microphone icon appears near your cursor. Start speaking.
Apple Dictation handles basic punctuation if you say it: "period," "comma," "question mark," "new paragraph." It also supports commands like "select all," "undo that," and "delete" for basic editing by voice.
When you're done, press the shortcut again or click the keyboard icon. The microphone icon disappears and your dictated text stays in place.
On-device vs server-based processing
On Apple Silicon Macs, Dictation runs entirely on-device by default. Your audio never leaves your Mac. This is a meaningful privacy improvement over the Intel era, when all dictation audio was sent to Apple's servers for processing.
Intel Macs still send audio to Apple's servers. If you're on an older Mac, be aware that your dictation is being processed in the cloud.
You can verify this in System Settings under Keyboard. The Dictation section will indicate whether processing happens on-device or requires a network connection.
Apple Dictation keyboard shortcuts
| Action | Shortcut |
|---|---|
| Start/stop dictation | Double-tap Fn (default) |
| Insert period | Say "period" or "full stop" |
| Insert comma | Say "comma" |
| New line | Say "new line" |
| New paragraph | Say "new paragraph" |
| Insert question mark | Say "question mark" |
| Insert exclamation | Say "exclamation mark" |
| Capitalize next word | Say "cap" |
| All caps next word | Say "all caps" |
Enhanced Dictation (and why it's gone)
Older macOS versions offered Enhanced Dictation, which downloaded a larger speech model for offline use and continuous dictation. Apple removed this feature in macOS Ventura and later, replacing it with the on-device processing built into Apple Silicon.
If you see guides referencing Enhanced Dictation, they're outdated. On any Mac running macOS 13 or later with Apple Silicon, you already have on-device processing. There's no separate "enhanced" toggle to enable.
Where Apple Dictation falls short
Apple Dictation handles basic tasks, but daily use exposes real limits. Understanding these helps you decide whether the built-in option is enough or whether a dedicated tool is worth the investment.
The 60-second timeout
Apple Dictation stops listening after approximately 60 seconds of continuous speech. There's no warning — it just stops. If you're mid-sentence on a long email or document, you have to re-activate it and pick up where you left off.
For quick messages and short notes, this doesn't matter. For anything longer than a paragraph, it's a constant interruption that breaks your flow.
Accuracy ceiling
Independent testing puts Apple Dictation's accuracy between 88% and 92% for clear, conversational English. That's roughly 1 error every 10 to 12 words. On a 200-word email, that's 16 to 22 words wrong.
In practice, most of those errors cluster around proper nouns, technical terms, and words that sound alike. Apple Dictation has no custom vocabulary support, so you can't teach it your company's product names, medical terminology, or legal jargon.
No smart formatting
Apple Dictation gives you raw transcription. Every filler word — "um," "uh," "like," "you know" — lands in your text. There's no automatic cleanup, no sentence restructuring, and no context-aware formatting.
Modern AI dictation tools strip filler words automatically, handle capitalization and punctuation more reliably, and in some cases adjust the tone of your text based on where you're typing. Apple's tool does none of this.
Limited app support
While Dictation technically works in most text fields, it struggles with certain apps. Some web-based editors (Notion, Figma's text tool, certain CMS platforms) don't receive dictation input reliably. Terminal doesn't support it well. And some password managers and secure text fields block it entirely.
Third-party speech to text apps for Mac
When Apple Dictation isn't enough, the Mac ecosystem has several dedicated tools that address its specific weaknesses. Each takes a different approach to accuracy, privacy, and workflow integration.
Blazing Transcribe
Blazing Transcribe runs in the macOS menu bar and types transcribed text directly into whatever app has focus. It uses a Parakeet neural network model optimized for Apple's Neural Engine, delivering ~530ms latency and a 2.5% word error rate — roughly 97.5% accuracy versus Apple Dictation's 88-92%.
The standout feature is always-on voice activity detection (VAD). Instead of pressing a shortcut to start dictating, Blazing Transcribe listens continuously and automatically transcribes when you speak. No button, no shortcut, no activation step. You talk, words appear.
Everything runs locally. No audio leaves your Mac. No cloud processing, no internet required, no privacy tradeoffs.
Price: $7/month with a free trial.
Best for: Mac users who want speech to text that works everywhere they type, with high accuracy, no timeout, and true hands-free operation.
Wispr Flow
Wispr Flow takes a different approach: it sends your audio to cloud servers and uses large language models to not just transcribe but reformat your speech. Dictation in Slack comes out casual. In email, it sounds professional. The AI adjusts tone based on context.
That intelligence comes with tradeoffs. Your audio goes to external servers, which matters if you handle confidential material. Latency is higher than local tools because of the network round trip. And the AI sometimes rewrites what you actually said into what it thinks you meant, which isn't always what you want.
Price: ~$19/month.
Best for: People who want polished output and don't mind cloud processing.
SuperWhisper
SuperWhisper runs OpenAI's Whisper model locally on your Mac. It's built for power users who want deep customization: multiple AI modes, custom prompts, LLM integration for text polishing, and fine-grained control over how transcription behaves in different contexts.
The flexibility is real, but so is the learning curve. Configuring modes, choosing models, and setting up prompts takes time. If you want something that works immediately out of the box, SuperWhisper might be more tool than you need. For a detailed breakdown, see our SuperWhisper review.
Price: $9.99/month, or $249.99 lifetime.
Best for: Tinkerers and power users who want maximum control over their dictation pipeline.
macOS Dictation (revisited)
For completeness: Apple Dictation costs nothing, requires no installation, and works well enough for occasional short dictation. If you send a few voice-typed messages per day and rarely dictate more than a sentence or two, it handles the job.
The moment you need extended dictation, custom vocabulary, or accuracy above 92%, a third-party tool pays for itself in saved correction time.
Comparison: speech to text tools for Mac
| Feature | Apple Dictation | Blazing Transcribe | Wispr Flow | SuperWhisper |
|---|---|---|---|---|
| Price | Free | $7/mo | ~$19/mo | $9.99/mo |
| Accuracy | 88-92% | ~97.5% (2.5% WER) | ~95% | ~95-97% |
| Latency | ~1-2s | ~530ms | ~1-3s (cloud) | ~1-2s |
| Processing | On-device (Apple Silicon) | On-device (Neural Engine) | Cloud | On-device |
| Timeout | ~60 seconds | None | None | None |
| Always-on mode | No | Yes (VAD) | No | No |
| Custom vocabulary | No | Yes | No | Via custom modes |
| Works system-wide | Most apps | All apps | All apps | All apps |
| Offline | Apple Silicon only | Yes | No | Yes |
| Smart formatting | No | Filler removal | AI rewriting | LLM polishing |
Tips for better speech to text accuracy on Mac
Regardless of which tool you use, these practices improve accuracy across the board.
Use an external microphone
Your MacBook's built-in microphone picks up fan noise, keyboard clicks, and room echo. A USB headset with a close-talk mic or a dedicated desk microphone reduces background noise and gives the transcription engine cleaner audio to work with.
You don't need expensive equipment. A $30 USB headset outperforms a MacBook's built-in mic for dictation. The gain is in proximity — a mic 2 inches from your mouth captures your voice clearly while rejecting ambient sound.
Speak naturally, not robotically
Early dictation software rewarded slow, deliberate enunciation. Modern AI models are trained on natural speech patterns. Speaking at your normal pace and cadence actually produces better results than artificially slowing down or over-pronouncing words.
Pause briefly between sentences instead of between words. The AI uses sentence-level context to resolve ambiguous words, so a natural sentence delivered at normal speed gives it more to work with than isolated words spoken one at a time.
Reduce background noise
Close windows, turn off fans or music, and avoid dictating in busy cafes or open offices. Even the best transcription models lose accuracy when competing with ambient sound.
If you can't control your environment, a directional microphone or a headset with active noise cancellation helps significantly. The goal is to increase the ratio of your voice to everything else.
Position matters
Sit up straight. Seriously. Slouching compresses your diaphragm and changes your vocal quality, which affects recognition. A clear, projected voice with consistent volume produces the cleanest transcription.
Keep a consistent distance from your microphone. Moving closer and farther away during dictation causes volume swings that confuse the automatic gain control on most mics.
Use custom vocabulary when available
If your work involves specialized terminology — medical terms, legal jargon, product names, technical acronyms — use a tool that supports custom vocabulary. AI dictation software with custom vocabulary can learn terms like "Kubernetes," "amoxicillin," or "voir dire" that generic models consistently miss.
Apple Dictation doesn't support this. Blazing Transcribe, SuperWhisper, and several other third-party tools do.
Speech to text for specific Mac workflows
Writing and content creation: Dictating a 1,000-word article takes about 7 minutes versus 20+ minutes typing. Always-on VAD tools like Blazing Transcribe are ideal here — you don't break your train of thought to press a shortcut, you just talk through your ideas and clean up afterward.
Email and messaging: A 50-word Slack message takes 20 seconds to dictate versus a minute to type. Across a full day of communication, the savings compound fast. Tools with smart formatting handle capitalization and punctuation so you skip the editing step.
Coding and development: Developers use speech to text for documentation, commit messages, code comments, and Slack replies — not for writing code itself. Dictating prose while your hands rest between coding sessions reduces total keyboard strain significantly.
Accessibility and RSI: For users with repetitive strain injury or carpal tunnel, speech to text isn't a productivity hack. It's a necessity. The key requirement is true hands-free typing software with always-on voice activity detection, so you never press a key to start dictation. Apple Dictation requires a shortcut to activate, which defeats the purpose. VAD-based tools eliminate that barrier.
Getting started: recommended setup
If you've never used speech to text on Mac, here's the fastest path to a working setup:
-
Start with Apple Dictation to get comfortable speaking instead of typing. Enable it in System Settings, then Keyboard, then Dictation. Use it for a few days for short messages and notes.
-
Notice the limits. You'll hit the 60-second timeout, see accuracy errors on proper nouns, and get annoyed at filler words in your text. That's normal.
-
Try a dedicated tool. Once you know speech to text fits your workflow, test a purpose-built app. Blazing Transcribe offers a free trial so you can compare the accuracy, speed, and always-on mode against Apple Dictation without committing.
-
Invest in a microphone. Even a $30 USB headset dramatically improves accuracy with any tool. This is the highest-ROI upgrade in the entire setup.
-
Build the habit. Speech to text becomes natural after about a week of consistent use. The first few days feel slower because you're thinking about the process. By day seven, you're just talking and text appears.
For a deeper comparison of every Mac dictation option available right now, see our guide to the best dictation app for Mac or our broader roundup of voice typing software.
Try Blazing Transcribe
If you want to go beyond Apple Dictation and use speech to text on Mac at its full potential, Blazing Transcribe is the fastest path. Local AI on your Mac's Neural Engine, ~530ms latency, 97.5% accuracy, and always-on voice detection that types wherever your cursor sits.
- Runs entirely on-device — fully private, no cloud
- Works in every Mac app: email, docs, Slack, browsers, code editors
- Always-on VAD — no shortcut needed, just talk
- $7/month with a free trial, no credit card required
Try Blazing Transcribe free at blazingfasttranscription.com
FAQ
How do I turn on speech to text on my Mac?
Open System Settings, click Keyboard in the sidebar, scroll to the Dictation section, and toggle it on. Choose your language and preferred keyboard shortcut (default is double-tap Fn). Once enabled, place your cursor in any text field and press the shortcut to start dictating. On Apple Silicon Macs, all processing happens on-device.
Is Mac speech to text accurate enough for professional use?
Apple's built-in Dictation achieves 88-92% accuracy, which means roughly 1 error every 10-12 words. For professional use — long emails, reports, documentation — that error rate creates significant editing overhead. Dedicated tools like Blazing Transcribe hit ~97.5% accuracy (2.5% word error rate), which is the threshold where most users find dictation faster than typing even after corrections.
Can I use speech to text offline on Mac?
Yes, if you have an Apple Silicon Mac (M1 or later). Apple Dictation processes speech on-device without an internet connection. Third-party tools like Blazing Transcribe and SuperWhisper also run fully offline using neural network models on the Mac's Neural Engine. Intel Macs send Apple Dictation audio to Apple's servers, so they need internet access for the built-in option.
What is the best speech to text app for Mac in 2026?
The best speech to text app for Mac depends on your priorities. For the highest accuracy with the least setup, Blazing Transcribe delivers 97.5% accuracy, ~530ms latency, and always-on voice detection for $7/month. For maximum customization and LLM-powered text polishing, SuperWhisper offers deep configuration options. For free and basic, Apple Dictation handles short bursts. See our full ranking of the best speech to text software for a detailed comparison.
How do I dictate in Microsoft Word on Mac?
To dictate in Microsoft Word on Mac, you have two options. First, use Apple Dictation: place your cursor in your Word document and press your dictation shortcut (double-tap Fn). Second, use Word's built-in dictation: click the Dictate button in the ribbon under the Home tab. Third-party system-wide tools like Blazing Transcribe work in Word automatically — just speak and text appears at your cursor. For more details, see our guide on voice typing in Microsoft Word.