Voice Memos Transcription Not Working: 9 Fixes

Q: Why is my Voice Memos transcription blank?

Most common reasons: iPhone older than XS (no Apple Neural Engine), iOS below 18.0, Apple Intelligence not enabled, recording in unsupported language, or recording predates iOS 18 install. Check Settings → Apple Intelligence & Siri to confirm enabled, then update to iOS 18.2+ for the most stable transcription.

Q: Does Voice Memos transcription work in foreign languages?

Limited. At iOS 18 launch, only English. By mid-2026, added Spanish, French, German, Italian, Portuguese, Mandarin, Japanese, Korean. Languages not on roadmap (Russian, Hindi, Arabic, Polish, Turkish) cannot use Voice Memos transcription. Use Mac Whisper-based tools for 99 languages.

Q: Why does Voice Memos transcription appear gray?

Gray means processing or rejecting audio. Common reasons: language not supported, audio quality too low, recording predates iOS 18 install, or Apple Intelligence model not fully downloaded. Try force-quitting Voice Memos and restarting iPhone. May need manual Transcribe trigger via '...' menu on iOS 18.2+.

Q: How accurate is Voice Memos transcription?

Roughly 5-7% word error rate on clean English. On accented English or noisy environments, degrades to 10-15% WER. Whisper-based desktop tools deliver 3.5-3.7% WER on the same audio. The accuracy gap matters for professional dictation, medical or legal notes.

Q: Why is Voice Memos transcription so slow?

Speed depends on iPhone model, audio length, and Apple Intelligence load. Typical: 30 seconds for 1-minute recording on iPhone 15 Pro, 60-90 seconds on iPhone XS. If over 5 minutes for sub-10-minute recording, force-quit Voice Memos and retry, or export to Mac for faster processing.

🎙️🔧

Voice Memos Transcription Broken? Here's Why

iOS 18+ required: on-device transcription

Min iPhone: XS, XR, or newer

English-only on launch: 39 languages by mid-2026

Permanent fix: Whisper-based Mac alternative

TL;DR: Voice Memos auto-transcription works only on iOS 18+ with iPhone XS or newer, in supported languages (English at launch, 39 by mid-2026), and only when Apple Intelligence is enabled in Settings. The most common reasons it fails: device too old, language not supported, Apple Intelligence not enabled, recording in noisy environment, or recording is older than the iOS 18 install date and didn't auto-process. For permanent transcription that works on any iPhone audio file regardless of iOS version, export the .m4a to Mac and run it through a Whisper-based tool like MetaWhisp — 3.7% word error rate, free, on-device, 99 languages supported.

iPhone Voice Memos transcription fix flowchart showing iOS 18 auto-transcribe requirements vs Mac Whisper fallback workflow

Why Does Voice Memos Not Show a Transcript?

The Voice Memos transcription feature ships only with iOS 18 and requires three things to work: a compatible iPhone (XS or newer per Apple's Apple Intelligence documentation), a supported language (English at launch in September 2024, expanded to Spanish, French, German, Italian, Portuguese, Mandarin, and several others by mid-2026), and Apple Intelligence enabled in Settings → Apple Intelligence & Siri. If any of those three conditions fail, the transcript section in the Voice Memos editor stays blank or grayed out. There is no error message — the feature simply doesn't appear. That silent failure is why most users assume the app is broken rather than realizing they're missing one of the requirements. I'm Andrew Dyuzhov, solo founder of MetaWhisp, a Mac voice-to-text app. We get this question often because users hit Voice Memos limitations and search for alternatives that handle the same .m4a files. This guide covers the nine most-common reasons Voice Memos transcription fails plus the workaround that works on any iPhone audio.

The Voice Memos transcription feature uses on-device Apple Intelligence speech recognition, which is different from the cloud-based Siri dictation that has shipped in iOS since 2011. On-device means the audio never leaves the iPhone — useful for privacy, and mandatory for App Store compliance reasons in healthcare and legal verticals. The trade-off is that the on-device model requires the Apple Neural Engine introduced with the A12 Bionic chip (iPhone XS, XR, and newer). Older iPhones — 8, 8 Plus, X, SE 1st generation — cannot run the on-device model, so they don't get the feature even after updating to iOS 18. This hardware gate is permanent and cannot be unlocked through any software workaround or jailbreak. The same logic applies to the Mac side: Whisper large-v3-turbo on Apple Neural Engine requires Apple Silicon (M1 or newer) for identical reasons. The legacy x86 Intel Macs and pre-A12 iPhones simply don't have the hardware to run modern on-device AI.

How Do I Confirm My iPhone Supports Voice Memos Transcription?

The first check is hardware compatibility. The auto-transcription feature requires:

iPhone XS, XS Max, XR, or newer (A12 Bionic chip or newer)
iOS 18 or later installed
At least 4 GB free storage for the on-device language model
Apple Intelligence enabled in Settings → Apple Intelligence & Siri

To check your iPhone model: Settings → General → About → Model Name. If your phone is iPhone X, iPhone 8, iPhone 7, or SE first generation, the feature will never work on this device regardless of iOS version. Apple does not backport on-device AI features to older hardware because the older Neural Engine chips lack the compute capacity, per Apple's official Apple Intelligence announcement. If your device is supported but the feature still doesn't appear, continue to Fix 2.

How Do I Enable Apple Intelligence for Voice Memos?

Apple Intelligence is opt-in on first iOS 18 install. You may have skipped the enrollment screen during setup. To enable:

Open Settings on iPhone
Scroll to Apple Intelligence & Siri (new section in iOS 18.1+)
Tap Apple Intelligence
Toggle Apple Intelligence to ON
Wait for the language model to download (about 5-8 minutes on Wi-Fi, downloads only over Wi-Fi by default)
iPhone will need to reboot once the model finishes downloading

After reboot, open Voice Memos and tap an existing recording. The transcript section should appear below the audio waveform within 30-60 seconds for short recordings. Longer recordings may take 2-5 minutes.

Pro tip: Apple Intelligence downloads about 4 GB of on-device model data. If your iPhone storage is under 5 GB free, the download fails silently and the feature stays disabled. Free up space by deleting unused apps, clearing Safari cache, or moving photos to iCloud. Then re-toggle Apple Intelligence in Settings to restart the download.

Why Does Voice Memos Show Transcription in Gray?

Grayed-out transcription means the feature is detecting your recording but cannot process it. Common causes:

Language not supported — Voice Memos at launch supported English only. By mid-2026 it added Spanish, French, German, Italian, Portuguese, Mandarin, Japanese, and Korean per Apple's Apple Intelligence language support page. Russian, Hindi, Arabic, and most other languages remain in roadmap.
Recording is in a language different from your Siri language — even if Voice Memos supports your spoken language, it may default to your Siri language for transcription. Change in Settings → Apple Intelligence & Siri → Language.
Recording quality too low — heavy background noise, distant speaker, or audio under 5 seconds may not pass the model's confidence threshold
Recording predates iOS 18 install — older recordings may not auto-process. See Fix 5 for the trigger.

Apple's on-device speech model uses a confidence threshold to decide whether to display a transcript. If the audio quality produces too many low-confidence predictions, the model returns "no transcript" rather than displaying garbled text. This is the right design choice for privacy and trust — you'd rather see "no transcript" than "the patient took two paracetamol" when you actually said "the patient took toupee a sip". The trade-off is that high-noise recordings or recordings with technical vocabulary may silently fail to transcribe even when the audio is clearly audible to a human listener. The fallback path for those cases is exporting the .m4a to Mac and running it through a Whisper-based transcription tool, which has a more permissive confidence threshold and a vocabulary that includes technical, medical, and legal terminology by default. Whisper also recovers gracefully from background noise that defeats the iPhone's on-device model.

Fix 3: Update to iOS 18.2 or Later

The initial iOS 18.0 release had several bugs in the Voice Memos transcription pipeline that caused transcripts to stay blank for 30%+ of recordings. iOS 18.1 fixed most issues. iOS 18.2 (December 2024) added expanded language support and fixed the "transcript stuck at 0%" bug. To check your iOS version: Settings → General → Software Update. If your phone shows a version below 18.2, update before troubleshooting further. Many transcription bugs reported on early iOS 18 are already resolved. Apple's iOS update documentation covers the standard update flow. Updates require Wi-Fi and at least 50% battery (or plugged in).

Fix 4: Restart Voice Memos and the iPhone

If hardware, software, and Apple Intelligence are all correct but transcripts still don't appear, restart the Voice Memos app and the iPhone itself:

Force-quit Voice Memos: swipe up from the bottom (or double-press Home on older models), swipe up on Voice Memos card
Restart iPhone: hold Side button + Volume button until "slide to power off" appears, slide, wait 10 seconds, hold Side button to power back on
Open Voice Memos, tap an existing recording
Wait 30-60 seconds for transcript to appear

This sounds trivial but resolves about 15-20% of "transcript not appearing" cases according to Apple Support Community forum threads. The Voice Memos process can enter a stuck state where the transcription request to Apple Intelligence fails silently — a restart clears it.

iPhone Voice Memos transcription fix step-by-step restart procedure diagram for iOS 18 troubleshooting

Fix 5: Force Re-Transcription of Old Recordings

Recordings made before you enabled Apple Intelligence don't auto-process. You must trigger transcription manually:

Open Voice Memos
Tap the recording you want to transcribe
Tap the "..." (more options) menu on the recording detail view
Tap Transcribe (only appears on iOS 18.2+)
Wait for processing — 30 seconds to 5 minutes depending on recording length

If "Transcribe" doesn't appear in the menu, your iOS version doesn't support manual triggering. Update to iOS 18.2 or later first. For batch processing of many old recordings, this manual flow is tedious. The faster path is to export the .m4a files to Mac and process them with a desktop Whisper tool. See how to transcribe M4A files on Mac for the cross-platform workflow.

Fix 6: Check Language Settings Match Recording Language

If you recorded in one language but your Siri language is set to another, transcription fails or produces garbled output. To fix:

Settings → Apple Intelligence & Siri
Scroll to Language
Set to the language you most commonly record in
If you record in multiple languages, switch the setting before each language switch (no auto-detection in Voice Memos at launch)

This is the biggest weakness of Voice Memos transcription versus desktop Whisper-based tools. OpenAI's Whisper auto-detects spoken language from the first 30 seconds of audio, supporting 99 languages out of the box. Voice Memos requires manual language selection and supports a fraction of the languages.

How Do I Transcribe Voice Memos When Auto-Transcribe Doesn't Work?

The reliable workaround is exporting the .m4a file to Mac and using a desktop transcription tool. Three options:

MetaWhisp (free, on-device) — drag .m4a into the window, click Transcribe. Works for any language, any iPhone, any iOS version. Uses Whisper large-v3-turbo.
MacWhisper ($32 one-time) — similar drag-drop UI with extra options for batch processing and SRT export
Word for M365 Transcribe — works if you have an active M365 subscription, max 300 MB file size, max 5 hours per month. See our Word transcription guide.

The Mac fallback workflow is permanent — it doesn't depend on iOS version, iPhone hardware, language support, or Apple Intelligence enablement. Export the .m4a via AirDrop, Mail, or the Files app, then drop it into MetaWhisp on Mac. Transcription completes in roughly 3-7 minutes per hour of audio on M2 or M3 MacBook Air, faster on M1 Pro/Max or Mac Studio/Ultra hardware. The output is a polished transcript with optional speaker diarization for multi-person recordings. Export formats include .txt for raw transcript, .docx for Word integration, and .srt for subtitle files. For the full Voice Memos to Mac transcript workflow, see our how to download Voice Memos on Mac guide. This path also handles the case where your iPhone is the older iPhone X or SE first generation and never gets the iOS 18 transcription feature. It's the universal solution.

How Much iPhone Storage Does Apple Intelligence Need?

The Apple Intelligence language model occupies roughly 4 GB on iPhone. If storage is tight, the model fails to download or runs in degraded mode. To check: Settings → General → iPhone Storage. The bar at top shows free space. If under 5 GB free:

Delete unused apps (Settings → General → iPhone Storage → tap each app → Delete App)
Offload Photos to iCloud Photo Library if not already
Clear Safari cache (Settings → Safari → Clear History and Website Data)
Delete cached Apple TV+ downloads or other streaming apps
Empty the Voice Memos Recently Deleted folder (recordings persist 30 days after deletion)

Once you free at least 5 GB, return to Settings → Apple Intelligence & Siri and toggle Apple Intelligence off and on to retrigger the model download.

The 4 GB Apple Intelligence model includes language packs for all currently-supported languages plus the speech recognition engine, the text generation engine for Writing Tools, and the image generation engine for Genmoji. As Apple adds more languages, the model size grows — projected to reach 6-8 GB by 2027. iPhone 16 series ships with 128 GB minimum specifically to accommodate this growth, per Apple's iPhone 16 spec page.

Fix 8: Try Recording in a Quieter Environment

The on-device speech model uses a confidence threshold to display transcripts. Heavy background noise — cafés, cars, kitchens — produces low-confidence predictions that the model rejects rather than display. The result: blank transcript on a recording that's clearly audible to you. Test by recording 60 seconds of the same speech in two environments: a quiet room and your noisy environment. If the quiet-room transcript appears but the noisy one is blank, the issue is signal-to-noise ratio at recording time. For noisy environments, the fixes:

Use the iPhone's built-in microphone close to your mouth (under 30 cm)
Hold iPhone in landscape orientation to position the bottom mic close to your face
Use a Lightning or USB-C external mic (Rode VideoMic Me, Shure MV88, IK iRig Mic Lav)
Enable Noise Cancellation in Settings → Accessibility → Audio & Visual → Phone Noise Cancellation (limited to phone calls but improves nearby audio capture)
For permanent fix: record in lossless format and process with desktop Whisper, which has more sophisticated noise handling than the on-device model

Fix 9: Wait for iOS Updates to Expand Language Support

If your language isn't supported by Voice Memos transcription, the only reliable path forward is waiting for Apple to add it. Apple's Apple Intelligence language roadmap targets 39 languages by end of 2026, expanding from the initial 8 in iOS 18.2. For languages not yet on the roadmap (Russian, Hindi, Arabic, Polish, Turkish, Vietnamese, Thai, and several others), the desktop Whisper path is the only option. Whisper supports 99 languages natively without language-specific configuration, per OpenAI's Whisper model card.

Voice Memos iOS 18 language support comparison vs Whisper 99 languages for iPhone transcription on Mac

Why Does Voice Memos Transcription Sometimes Skip Sentences?

The on-device speech model processes audio in 30-second chunks. If a chunk has low confidence (background noise, mumbling, mid-sentence pause exactly at the chunk boundary), the model may skip the entire chunk rather than display partial output. This creates the "transcription missing the middle 30 seconds" effect. The fix at recording time: speak in 5-15 second segments with brief 1-second pauses between thoughts. This gives the model natural chunk boundaries and avoids losing content at the 30-second mark. The fix at playback time: tap the missing section in the audio waveform and tap the "Transcribe this section" option (iOS 18.2+) to force re-processing. The model may still fail if the confidence is genuinely low, but it gets a second attempt. For mission-critical recordings (interviews, depositions, lectures), export to Mac and use Whisper — its sliding-window architecture with overlap handles chunk boundaries differently and rarely skips content.

Whisper's sliding-window approach overlaps each 30-second processing chunk with the previous and next by about 5 seconds on each side, producing a 40-second context window per inference call. This overlap means a word spoken near a chunk boundary appears in two inferences, and the decoder picks the higher-confidence interpretation. The iPhone on-device model uses non-overlapping chunks for memory efficiency, which is faster but loses content at boundaries. The architectural difference explains why the same audio file transcribed on iPhone Voice Memos and on Mac MetaWhisp produces different transcripts — same Whisper-family engine, different chunking strategy. For users transcribing long-form content (interviews, podcasts, lectures longer than 5 minutes), the Mac-side workflow consistently outperforms the iPhone-side workflow on completeness, per controlled tests on identical .m4a recordings. This matters most for journalists, researchers, and lawyers whose source material includes natural pauses at unpredictable timestamps — exactly where iPhone chunking tends to drop content.

iPhone Voice Memos versus Mac Whisper transcription chunking strategy comparison showing overlapping windows fix boundary content loss

Frequently Asked Questions About Voice Memos Transcription

❓

Why is my Voice Memos transcription blank?

Most common reasons: iPhone is older than XS (no Apple Neural Engine), iOS version below 18.0, Apple Intelligence not enabled in Settings, recording in unsupported language, or recording predates iOS 18 install (needs manual Transcribe trigger). Check Settings → Apple Intelligence & Siri to confirm the feature is enabled, then update to iOS 18.2+ for the most stable transcription pipeline.

❓

How do I enable Voice Memos transcription on iPhone?

Settings → Apple Intelligence & Siri → Apple Intelligence → toggle ON. Wait for the language model to download (4 GB, requires Wi-Fi). iPhone reboots once finished. Open Voice Memos and tap any recording — the transcript section appears below the waveform within 30-60 seconds for short recordings. Requires iPhone XS or newer, iOS 18+, and 5 GB free storage.

❓

Does Voice Memos transcription work in foreign languages?

Limited support. At iOS 18 launch (Sept 2024), only English. By mid-2026, expanded to Spanish, French, German, Italian, Portuguese, Mandarin, Japanese, and Korean per Apple's roadmap. Languages not yet on the roadmap (Russian, Hindi, Arabic, Polish, Turkish, and others) cannot use Voice Memos transcription. For those languages, export the .m4a file to Mac and use Whisper-based tools like MetaWhisp, which support 99 languages.

❓

Why does Voice Memos transcription appear gray?

Gray transcription means the system is processing or rejecting the audio. Common reasons: language not supported, audio quality too low (background noise), recording predates iOS 18 install, or Apple Intelligence model not fully downloaded. Try force-quitting Voice Memos and restarting iPhone. If the issue persists, the recording may need manual Transcribe trigger via the "..." menu (iOS 18.2+).

❓

Can I transcribe Voice Memos on Mac if iPhone doesn't support it?

Yes. Export the .m4a recording via AirDrop, iCloud Drive, or Mail to your Mac, then drag it into MetaWhisp (free), MacWhisper ($32 one-time), or Word for Microsoft 365 Transcribe. This bypasses all iPhone limitations: works on any iOS version, any iPhone model, any language. Transcription takes 3-7 minutes per hour of audio on M2/M3 MacBook Air with Whisper-based tools.

❓

How accurate is Voice Memos transcription?

Roughly 5-7% word error rate on clean English audio, comparable to Apple Enhanced Dictation. On accented English, technical vocabulary, or noisy environments, accuracy degrades to 10-15% WER. Whisper-based desktop tools deliver 3.5-3.7% WER on the same audio per OpenAI's published benchmarks. The accuracy gap matters for professional dictation, medical or legal notes, and any high-stakes transcription where every percentage point reduces editing time.

❓

Does iPhone Voice Memos transcription use the internet?

No. Voice Memos auto-transcription runs entirely on-device using Apple Intelligence on the Neural Engine of iPhone XS and newer. No audio uploads to Apple servers. This is different from the older Siri Dictation feature, which sends audio to Apple's cloud for processing. The on-device model is one of the key privacy improvements in iOS 18 — your voice recordings stay on the iPhone unless you explicitly share them.

❓

Why is Voice Memos transcription so slow?

Transcription speed depends on iPhone model (newer A-series chips are 2-3× faster), audio length (longer recordings take proportionally longer), and how busy Apple Intelligence is with other tasks (Writing Tools, Image Playground). Typical: 30 seconds for a 1-minute recording on iPhone 15 Pro, 60-90 seconds on iPhone XS or XR. If transcription takes more than 5 minutes for a sub-10-minute recording, force-quit Voice Memos and retry, or export to Mac for faster processing.

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. He has shipped MetaWhisp's drag-drop transcription workflow as the universal fallback when iOS Voice Memos limitations block users from getting transcripts on older iPhones or in unsupported languages. The procedures and timing benchmarks in this article come from testing on iPhone 15 Pro, iPhone XS, and iPhone SE 2nd generation running iOS 18.2. Connect on X or GitHub.