Apple Intelligence Dictation Issues (2026)

Q: Why does Apple Dictation stop working after macOS update?

Most common cause: macOS updates reset Dictation switch to off. Open System Settings → Keyboard → Dictation and toggle on. Other causes: language pack corruption (re-download by switching languages), microphone permission reset (Privacy & Security → Microphone), Apple Intelligence model partial download (check Apple Intelligence & Siri settings).

Q: Did macOS Sequoia 15.2 fix Apple Dictation?

Partially. 15.2 fixed several regressions from 15.0/15.1 including reliability on Apple Silicon, microphone permission handling, language pack download stability. Architectural limits unchanged: silence cutoffs, no custom vocabulary, no accuracy fix on technical terms. Users still hitting limits typically benefit from third-party Whisper apps.

Q: Can I add custom vocabulary to Apple Dictation?

No. Apple Dictation uses fixed vocabulary in on-device model. No way to add brand names, technical terms, proper nouns. Workarounds: Text Replacements (Keyboard → Text Replacements) for auto-correction, or switch to Whisper-based apps like MacWhisper, MetaWhisp, SuperWhisper which support custom vocabulary lists.

Q: Why does Apple Dictation cut off after pauses?

Apple Dictation has built-in silence cutoffs — after few seconds of silence, session ends. Architecturally baked in for processing management. For long-form dictation with natural pauses, cutoff interrupts flow. Third-party Whisper apps handle longer silences using voice activity detection to ignore silence without ending session.

Q: What's the best alternative to Apple Dictation on Mac?

Free on-device no telemetry: MetaWhisp (Whisper large-v3-turbo on Apple Neural Engine, free, 99 languages, custom vocabulary). Paid one-time: MacWhisper ($29). Maximum customization: SuperWhisper. All run Whisper locally on Mac, none have silence cutoff or update-breakage issues of Apple Dictation, all support custom vocabulary.

🍎🎙️

Apple Intelligence Dictation Issues

Top cause: macOS update reset Dictation switch

15.2 fixed: Several regressions

Lasting: Silence cutoffs, no custom vocab

Workaround: Third-party Whisper apps

TL;DR: Apple Intelligence dictation breaks after macOS updates with predictable failure modes. The single most common cause is that minor updates reset the Dictation switch to off — re-enabling in System Settings → Keyboard → Dictation fixes it instantly. macOS Sequoia 15.2 addressed several earlier regressions per Apple's official guidance, but architectural limits remain: silence cutoffs, no custom vocabulary, accuracy regressions across releases, and language-pack corruption requiring re-download. For users who hit these limits repeatedly, third-party Whisper-based apps like MetaWhisp, MacWhisper, and SuperWhisper offer alternatives that work consistently across macOS updates because they bundle their own transcription models rather than relying on Apple's system-level service.

Flowchart diagram showing Apple Intelligence dictation failure diagnosis paths after macOS update for Mac users with three common causes and fixes

Why Does Apple Intelligence Dictation Break After macOS Updates?

The pattern is documented across years of macOS releases and confirmed in Apple Community discussion threads. macOS updates — both major releases and point updates — can cause four distinct dictation failures:

Dictation switch silently resets to off — Most common. Apple's update process re-applies system defaults, sometimes including the Dictation enable/disable toggle. The switch ends up off after a reboot, dictation appears to stop working.
Language pack partial download or corruption — Apple downloads language packs separately from the macOS installer. If a previous download was interrupted or corrupted, the on-device dictation model fails silently when you try to use it.
Apple Intelligence model partial download — Apple Intelligence has its own model files that download in the background over hours or days. If your Mac was offline during the download window, the model may be partial. Dictation features that depend on Apple Intelligence may not work until the download completes.
Microphone or Accessibility permission revoked — Updates occasionally reset privacy permissions. If Dictation lost access to your microphone, it fails silently with no clear error message.

The reason these aren't fixed permanently in updates is that they're not bugs in the traditional sense. They're emergent behaviors from how macOS handles system defaults, downloads, and privacy resets during updates. Apple has reduced the frequency over time but the pattern recurs.

The architectural reason Apple Dictation breaks after updates while third-party Whisper apps don't is the dependency structure. Apple Dictation depends on system-level services (the dictation daemon), system-level language packs (downloaded separately from the OS), and system-level permissions (which can reset). When any one of these breaks, dictation breaks. Third-party Whisper apps like MetaWhisp bundle the Whisper model with the app itself, request microphone permission once at install, and don't depend on background language pack downloads. The dependency chain is shorter and more local, so update-induced breakage is rarer. This is the trade-off: Apple Dictation is built into macOS but tied to its update cycle; third-party tools are extra installs but more update-resilient.

What Did macOS Sequoia 15.2 Fix for Dictation?

Per Apple's release notes and confirmed by user reports on Apple Community, macOS Sequoia 15.2 addressed several dictation regressions that had appeared in 15.0 and 15.1:

Improved reliability on Apple Silicon Macs — Issue where dictation would silently fail after periods of inactivity
Better handling of microphone permissions — Updates no longer reset microphone access for Dictation in most cases
Language pack download stability — Reduced (not eliminated) the rate of partial-download corruption
Apple Intelligence integration polish — Smoother handoff between Apple Intelligence Writing Tools and Dictation

What 15.2 did NOT fix:

Silence cutoff behavior (Apple Dictation still ends dictation sessions after extended pauses)
Lack of custom vocabulary (no way to add brand names, technical terms, or proper nouns)
Accuracy regressions on specific accents or audio conditions
Inability to use Dictation in some third-party text fields (longstanding compatibility issue)
Memory/CPU usage spikes on older Apple Silicon Macs during Enhanced Dictation operation

For users still hitting issues after 15.2, the diagnostic process below covers the common remaining causes.

How Do I Fix Apple Dictation That Stopped Working After Update?

The sequence from most-common to least-common cause:

Fix 1: Re-enable Dictation in System Settings

Open System Settings (formerly System Preferences)
Navigate to Keyboard → Dictation
If the Dictation switch is off, turn it on
If you're prompted to download a language pack, accept
Test dictation with the keyboard shortcut (default: press and hold the Globe/microphone key, or Function key twice depending on settings)

If this fixes it, the cause was the update resetting the switch. This is the most common cause and the easiest fix.

Fix 2: Re-download the Language Pack

System Settings → Keyboard → Dictation
Note your current Dictation Language
Change to a different language temporarily
Switch back to your original language — macOS will re-download the pack
Wait for download to complete (visible in download manager)
Test dictation

This forces macOS to re-download the on-device language model, which fixes corruption-related failures.

Fix 3: Reset Microphone Permission

System Settings → Privacy & Security → Microphone
Find Dictation in the list (may appear as "System Services" or similar)
Toggle microphone access off, then on
If asked, restart the Mac after toggling
Test dictation

This resets the privacy permission grant, sometimes needed after major updates that change permission models.

Fix 4: Check Apple Intelligence Setup (if applicable)

System Settings → Apple Intelligence & Siri
Verify Apple Intelligence is enabled (requires M-series Mac and supported region)
Check Download Status for any pending model downloads
Wait for downloads to complete before expecting Apple Intelligence-dependent dictation features

For users in regions without Apple Intelligence support, this step doesn't apply — standard Dictation should still work without Apple Intelligence features.

Fix 5: Reset Dictation Cache (Advanced)

Quit any apps using dictation
Open Terminal
Run: defaults delete com.apple.assistant.support
Restart your Mac
Re-enable Dictation in System Settings (it will need fresh setup)

This is a last-resort fix that clears Dictation's local cache. Use only if Fixes 1-4 didn't work.

Five step diagnostic flowchart for fixing Apple Dictation on Mac showing relative frequency of each cause from most common toggle reset to least common cache clear

Why Doesn't Apple Dictation Support Custom Vocabulary?

This is a longstanding architectural decision that hasn't changed in macOS Sequoia. Apple Dictation uses a fixed vocabulary trained into the on-device model. Users cannot add custom terms, brand names, technical vocabulary, or proper nouns. The practical consequence:

Dictating "MetaWhisp" produces "meta whisper"
Dictating "Kubernetes" sometimes works, sometimes produces "Cuban etis"
Names of less-common people, places, or products get phonetic substitutions
Internal jargon, code identifiers, scientific terms transcribe inconsistently

For users who need accurate transcription of specific vocabulary, the options are:

Manual correction after dictation (works but breaks flow)
Use TextReplacement entries to fix common errors automatically (System Settings → Keyboard → Text Replacements)
Switch to a third-party Whisper-based app with custom vocabulary support
Use Voice Control (Apple's accessibility feature) with custom commands

Whisper-based apps like MacWhisper and SuperWhisper expose custom vocabulary settings. The user provides a list of terms, and the app biases transcription toward those terms. This is the simplest way to handle technical vocabulary without manual correction.

When Should I Switch from Apple Dictation to a Third-Party Tool?

The decision threshold is when the limitations affect your daily workflow:

You dictate technical or domain-specific vocabulary — Custom vocabulary support is a hard win. MacWhisper, MetaWhisp, SuperWhisper all support custom term lists.
You dictate long-form content with natural pauses — Apple Dictation's silence cutoffs interrupt the flow. Third-party tools handle longer silences gracefully.
You hit accuracy issues that don't improve across updates — Whisper large-v3-turbo running on Apple Neural Engine consistently outperforms Apple Dictation on accented English and noisy audio.
You need transcription in 99 languages with auto-detection — Whisper's multilingual capability exceeds Apple Dictation's language coverage and switching UX.
You want stable behavior across macOS updates — Bundled-model apps don't have the same update-induced breakage pattern as system-level Dictation.

For users whose dictation is occasional and works fine when it works, sticking with Apple Dictation is reasonable. For users hitting limitations repeatedly, the migration to a Whisper-based app pays for itself in the first week.

The migration cost from Apple Dictation to a third-party Whisper app is small in 2026. Apps like MetaWhisp, MacWhisper, and SuperWhisper install in five minutes, request the same Microphone and Accessibility permissions Apple Dictation needs, and integrate with the same global hotkey workflow most users prefer. The transcript ends up in the same place — the active text field or clipboard. The user experience is similar enough that the transition feels like a configuration change rather than a workflow rebuild. The accuracy and feature upgrades show up immediately on the first dictation session. For most users who've been frustrated by Apple Dictation for years, the switch is a quick fix to a chronic problem rather than a major project.

Does Apple Intelligence Make Mac Dictation Better or Worse?

Mixed. Apple Intelligence on supported Macs adds three dictation-relevant features:

Writing Tools — Post-dictation rewriting, proofreading, summarization. Useful for cleaning up dictated drafts.
Smart Reply suggestions — Faster response composition in Mail and Messages, partly driven by dictation input.
Improved Siri context awareness — Siri remembers context across requests, including dictated input.

What Apple Intelligence does not fix:

Custom vocabulary (still no way to add terms)
Silence cutoff behavior (architectural limit unchanged)
Accuracy on accented English, noisy audio, or technical vocabulary (model is the same)
Update-induced breakage patterns (system-level dependency chain unchanged)

For users on supported Macs in supported regions, Apple Intelligence's Writing Tools are genuinely useful as a post-dictation cleanup layer. The underlying dictation model and its limitations are unchanged. For users in unsupported regions or on Intel Macs, Apple Intelligence isn't available, but standard Dictation still works (with the limitations described above).

What's the MetaWhisp Take on Apple Dictation?

I'm Andrew Dyuzhov, founder of MetaWhisp. For users whose Apple Dictation works fine, I don't think they need MetaWhisp. Apple Dictation is built in, free, and adequate for casual personal dictation. MetaWhisp exists for the users who hit Apple Dictation's limitations:

Users dictating technical content who lose 10-15% of their dictation time to manual correction of jargon
Users in regulated industries who can't tell whether Apple Dictation routes audio to Apple's cloud (Enhanced Dictation runs on-device but standard Dictation may not)
Users dictating in non-English languages where Apple Dictation accuracy lags Whisper
Users who got tired of Apple Dictation breaking after every other macOS update

MetaWhisp runs Whisper large-v3-turbo on Apple Neural Engine, supports 99 languages with auto-detection, never requires Screen Recording permission, and works the same across macOS update cycles because the model is bundled with the app rather than pulled from system-level services. Free. The recommendation for new Mac users: try Apple Dictation first. If it works for you, great. If you hit the limitations above repeatedly, the migration to a Whisper-based app is a small lift with meaningful quality-of-life improvement.

Frequently Asked Questions About Apple Intelligence Dictation

❓

Why does Apple Dictation stop working after macOS update?

The most common cause is that macOS updates reset the Dictation switch to off. Open System Settings → Keyboard → Dictation and toggle it back on. Other causes include language pack corruption (re-download via switching languages back and forth), microphone permission reset (Privacy & Security → Microphone), and Apple Intelligence model partial download (check Apple Intelligence & Siri settings).

❓

Did macOS Sequoia 15.2 fix Apple Dictation?

Partially. 15.2 fixed several regressions from 15.0 and 15.1 including reliability on Apple Silicon, microphone permission handling, and language pack download stability. Architectural limits remain unchanged: silence cutoffs, no custom vocabulary support, no way to fix accuracy on technical terms. Users still hitting these limits typically benefit from third-party Whisper-based apps.

❓

Can I add custom vocabulary to Apple Dictation?

No. Apple Dictation uses a fixed vocabulary trained into the on-device model. No way to add brand names, technical terms, or proper nouns. Workarounds: use Text Replacements (Keyboard → Text Replacements) to auto-correct common errors after dictation, or switch to a Whisper-based app like MacWhisper, MetaWhisp, or SuperWhisper that supports custom vocabulary lists.

❓

Does Apple Intelligence improve dictation accuracy?

Not directly. Apple Intelligence adds Writing Tools for post-dictation rewriting/proofreading, Smart Reply suggestions, and improved Siri context. The underlying dictation model is unchanged, so accuracy on technical vocabulary, accents, and noisy audio is the same as without Apple Intelligence. For accuracy improvement, third-party Whisper-based apps using Whisper large-v3 or larger models perform better.

❓

Why does Apple Dictation cut off after pauses?

Apple Dictation has built-in silence cutoffs — after a few seconds of silence, the dictation session ends. This is architecturally baked in to manage processing. For long-form dictation with natural pauses, the cutoff interrupts flow. Third-party Whisper-based apps handle longer silences gracefully by using voice activity detection to ignore silence without ending the session.

❓

Is Enhanced Dictation different from regular Dictation on Mac?

Enhanced Dictation runs on-device on Apple Silicon Macs, processing audio locally without internet connectivity. Regular Dictation may route to Apple's servers depending on configuration. For privacy-sensitive content, ensure Enhanced Dictation is enabled (System Settings → Keyboard → Dictation, look for "use enhanced dictation" or equivalent). On modern Macs with Apple Silicon this is generally the default.

❓

What's the best alternative to Apple Dictation on Mac?

For free on-device with no telemetry: MetaWhisp (Whisper large-v3-turbo on Apple Neural Engine, free, supports 99 languages, custom vocabulary). For paid one-time: MacWhisper ($29). For maximum customization: SuperWhisper. All run Whisper locally on Mac, none have the silence cutoff or update-breakage issues of Apple Dictation, all support custom vocabulary.

About the Author

Andrew Dyuzhov is the solo founder and CEO of MetaWhisp, a free on-device voice-to-text app for macOS that runs Whisper large-v3-turbo on Apple Neural Engine. MetaWhisp was built in part to address the recurring Apple Dictation issues that resurface with each macOS update — bundling the transcription model with the app rather than depending on system-level services produces more stable behavior across update cycles. Connect on X or GitHub.