Mac Dictation Not Working? Here's Why (And What Actually Works)

Let me guess. You updated macOS, opened a document, hit that little microphone button, and... nothing. Or worse — it picked up your voice but turned "schedule a meeting with the design team" into "schedule a meanie with the sign teen."

You're not alone. "Mac dictation not working" is one of the most searched Apple-related frustrations on the internet. And it's been that way for years.

I've been down this rabbit hole myself. Toggled settings, restarted Siri, reset the SMC, sacrificed a USB-C dongle to the Apple gods. Eventually I stopped trying to fix Apple's dictation and built something better. But we'll get to that.

First, let's actually try to fix your problem.

The Usual Suspects: Why Mac Dictation Breaks

It Broke After a macOS Update

This is the #1 reason people search for this. Every major macOS release — and plenty of minor ones — seems to break dictation for a chunk of users. The microphone stops responding, the feature simply won't activate, or accuracy nosedives.

Why? Because Apple's dictation relies on a web of system services: Siri, keyboard settings, microphone permissions, and (for Enhanced Dictation) on-device models that sometimes get corrupted during updates.

Quick fixes to try:

1. Toggle it off and on. System Settings → Keyboard → Dictation. Turn it off, wait 10 seconds, turn it back on.
2. Check your microphone permissions. System Settings → Privacy & Security → Microphone. Make sure the app you're dictating into has permission.
3. Reset Siri. Disable Siri entirely, restart your Mac, then re-enable both Siri and Dictation.
4. Delete the speech recognition cache. Open Terminal and run: rm -rf ~/Library/Caches/com.apple.SpeechRecognitionCore then restart.
5. Check your internet connection. Standard Mac dictation (not Enhanced) requires an active internet connection. Yes, really.

It Requires the Internet (No, Seriously)

Here's the thing most people don't realize: unless you've specifically enabled Enhanced Dictation (and it's available for your language), every word you speak gets sent to Apple's servers for processing.

No Wi-Fi? No dictation. Spotty coffee shop internet? Garbled results. On a plane? Forget it.

This is 2026. We have AI models that run entirely on a phone chip. The fact that Apple still ships cloud-dependent dictation as the default is... a choice.

The Accuracy Is Just Bad

Even when Mac dictation technically works, the accuracy often doesn't. Technical terms get mangled. Names become creative fiction. Anything beyond simple English sentences is a coin flip.

The core issue: Apple's dictation model isn't Whisper. It's not even close to Whisper. OpenAI's Whisper model — which powers most third-party dictation apps now — is dramatically more accurate, especially with accents, technical vocabulary, and natural speech patterns.

Apple's model was fine in 2015. In 2026, it feels like using MapQuest when Google Maps exists.

Enhanced Dictation Has Its Own Problems

Apple introduced Enhanced Dictation as the on-device alternative. Great idea in theory. In practice:

• It's not available for all languages
• The on-device model is less accurate than the cloud version for many users
• It occasionally just... stops working, requiring you to re-download the language model
• It doesn't support continuous dictation well — it times out

If Enhanced Dictation worked perfectly, I wouldn't have built TAWK. But here we are.

The Real Problem: Apple Doesn't Care About Dictation

I'm going to say something that might sound harsh: dictation isn't a priority for Apple.

Think about it. When was the last time Apple showed off dictation improvements at WWDC? When did they last blog about speech recognition accuracy? They've poured resources into Siri (which still can't set two timers reliably) while leaving the core dictation feature to rot.

Meanwhile, OpenAI released Whisper as an open-source model in 2022, and within months, a dozen indie developers built dictation apps that absolutely smoke Apple's built-in offering.

The gap isn't closing. It's widening.

What Actually Works: Your Real Options

If you're tired of fighting with Mac dictation, here are your actual alternatives. I'll be honest about all of them, including the one I built.

Option 1: Keep Troubleshooting Apple's Dictation

If you just need basic dictation for occasional use and don't want to install anything, the troubleshooting steps above might get you back on track. Apple's dictation is fine for short bursts of simple text.

Best for: Casual users who dictate a sentence or two at a time.

Option 2: Use a Whisper-Based Dictation App

This is where the real accuracy lives. Apps that use OpenAI's Whisper model running locally on your Mac deliver significantly better results than Apple's built-in dictation. No internet required, no cloud processing, much better accuracy.

Several options exist in this space — I wrote a full comparison of the best voice-to-text apps for Mac if you want the deep dive. The short version:

• Subscription apps like Wispr Flow ($144/yr) and Superwhisper (~$100/yr) offer AI text cleanup and advanced features, but you're paying monthly forever.
• Free options like FluidVoice exist but can be rough around the edges.
• TAWK (that's the one I built) costs $29 once. No subscription, no account, no cloud. Press a hotkey, talk, text appears at your cursor.

Option 3: Use a Cloud-Based Service

Google's speech-to-text API, Amazon Transcribe, and similar services offer great accuracy. But they require developer setup, ongoing API costs, and your audio goes to someone else's servers. Not exactly plug-and-play for writing an email.

Why I Stopped Fixing Apple's Dictation and Built TAWK

I'm a developer. I spent way too much time troubleshooting Mac dictation — resetting caches, toggling settings, restarting services. Every macOS update was a dice roll.

Then I discovered Whisper. OpenAI's speech recognition model is absurdly good. It handles accents, technical jargon, fast speech, and mumbling better than anything I'd used before. And it runs locally — no internet, no cloud, no privacy concerns.

The problem? Using Whisper meant running Python scripts in Terminal. Not exactly a workflow for writing Slack messages.

So I wrapped it in the simplest possible Mac app. Menu bar icon, global hotkey, text appears at your cursor. That's TAWK.

No account to create. No subscription to manage. No audio leaving your machine. $29 once, and it's yours forever. Read more about why local Whisper processing matters for privacy.

Quick Troubleshooting Checklist

Before you go, here's the condensed version for the "I just need this to work right now" crowd:

1. ✓ Toggle Dictation off/on in System Settings → Keyboard
2. ✓ Restart your Mac (yes, really)
3. ✓ Check microphone permissions for your app
4. ✓ Verify internet connection (standard dictation needs it)
5. ✓ Reset Siri, then re-enable Dictation
6. ✓ Delete speech recognition cache via Terminal
7. ✓ Try a different microphone input (Settings → Sound → Input)
8. ✓ If using Enhanced Dictation, re-download the language model
9. ✓ Check if a third-party app is hogging the microphone
10. ✓ Consider that maybe it's not you — it's Apple's dictation. Try a Whisper-based alternative.