Skip to content
Voice Memos

How do I transcribe voice memos on iPhone for free?

By ·

Free voice memo transcription on iPhone in 2026 has gotten dramatically better with iOS 18 and especially iOS 26. Here's what's free and what isn't.

Apple Voice Memos (iOS 18+) — free, on-device, decent

If you're on iOS 18 or later, every voice memo you record gets auto-transcribed on-device using Apple's speech recognition. To view:

  • Open Voice Memos app.
  • Tap any recording.
  • Tap the transcript icon (looks like text lines, near the playback controls).
  • The transcript appears synced to playback — tap any word to jump to it.

Quality is solid for clear speech in English. Mediocre for accents, multiple speakers, or background noise.

Apple Notes voice recording (iOS 17+) — free, on-device

Notes has a voice recording feature with auto-transcription. The transcription quality is similar to Voice Memos but the transcripts are searchable within Notes itself, which is a big advantage.

Nemos (free) — on-device + searchable + multi-device

Nemos records voice notes and runs Apple's on-device Speech framework for transcription. The advantage over Voice Memos is that transcripts are fully searchable across your whole library, syncable to iPad/Mac/Apple Watch, and you can tag/folder recordings. Free tier covers unlimited recordings.

Whisper-based open-source tools (free, off-device)

For long recordings (1+ hour meetings) or non-English audio, OpenAI's open-source Whisper model running locally via apps like MacWhisper (Mac) or AudioPen's free tier (iOS) gives near-perfect accuracy. Trade-off: setup is non-trivial on iPhone.

Free vs paid in 2026 — what's worth paying for?

  • Speaker diarization (telling speakers apart in a meeting) — paid features in Otter, Granola, Notta.
  • Real-time transcription during the call — Otter and Granola, $15-25/mo.
  • Punctuation and paragraph cleanup — most free tools dump everything in one paragraph.

For everyday voice memos, the free Apple/Nemos combo is enough for 95% of users. Save the paid tools for actual meetings.

## Why this question gets asked so often

Voice memo transcription was a $20+/month feature until iOS 18 shipped on-device transcription in September 2024 — a watershed moment that wiped out the value proposition of cheaper paid transcription apps. The pre-2024 landscape included Otter ($16.99/mo), Rev ($0.25/min), Trint ($60/mo), Descript ($24/mo), and several less-known apps that charged for what Apple now does for free. Google search volume for "free voice transcription iPhone" jumped 480% between September 2024 and January 2025 as users discovered the new feature. The question keeps trending because Apple buried the transcript button in a non-obvious place (you have to tap the recording, then tap the transcript icon — the discoverability is poor). App Store reviews of Voice Memos consistently mention surprise: "Wait, this was always free?" The deeper story is that on-device transcription is genuinely difficult — Whisper-quality output requires 2-4 GB of model weights and significant Neural Engine time, and Apple invested heavily to make it work on consumer hardware.

## The deeper story

Apple's voice transcription pipeline uses a tiered approach: lightweight wake-word detection (Always-On Processor), audio recording (Audio framework), and transcription (Speech framework, now backed by a 1.2B parameter speech model on iOS 26+ devices). The transcription model is roughly equivalent in quality to OpenAI's Whisper "small" model but runs on-device in 0.4x real-time (a 60-second recording transcribes in ~24 seconds). For comparison, cloud-based services like Otter use Whisper "large" variants that achieve 0.1x real-time but require server round-trips. The accuracy gap is most noticeable on accented English (Apple's model trained heavily on US/UK/Australian/Indian English; less so on Nigerian, Pakistani, or Singaporean English) and on overlapping speakers. The 2024 Tiago Forte BASB workflow calls voice memos "the highest-leverage capture format" because speaking is 3x faster than typing — transcription democratizes that across devices and use cases.

## Edge cases and gotchas

  • Apple Watch dictation vs full recording: dictation captures text-only; full recording captures audio AND transcribes. Different use cases.
  • Lossless audio recordings: don't transcribe better than standard quality — speech is well below the audible frequency range that lossless preserves.
  • Bilingual recordings: Apple's Speech framework only transcribes the dominant language. Code-switching content (e.g., Spanglish) loses the secondary language.
  • Background music: heavy music can confuse the model, especially when lyrics overlap speech.
  • Phone call recording: iOS 18+ lets you record calls with on-device transcription, but the other party must consent (a tone plays). Transcripts are stored in the Phone app, not Voice Memos.
  • Whisper-quality fans: MacWhisper Pro ($29 one-time) gives near-perfect on-device transcription on Mac, useful for long recordings.

## What competitors say

Otter ($16.99/mo) was the category leader for years; iOS 18 transcription has eaten significant market share. Their differentiator now is real-time meeting transcription with speaker diarization. Granola ($25/mo) targets the meeting-summary niche with LLM post-processing. Notta ($14.99/mo) offers similar features with translation. MacWhisper runs OpenAI Whisper locally on Mac — best for long-form journalism and podcast workflows. Apple Notes matches Voice Memos' on-device transcription, with the added benefit of searchability within Notes. Bear doesn't natively record audio. Notion doesn't natively record audio (requires third-party integration). Obsidian uses community plugins for audio capture. Nemos layers semantic search on top of Apple's Speech framework, so finding voice memos by concept (not just keyword) works.

## The 2026 verdict

If you're on iOS 18 or later, free on-device transcription via Voice Memos or Apple Notes is genuinely excellent for everyday use — buy a paid tool only if you need speaker diarization, real-time transcription during meetings, or specialized accuracy for professional content. The category leaders (Otter, Granola, Notta) all face the same pressure: Apple's free tier now does 85% of what they charge $15-25/mo for. Expect the meeting-transcription space to consolidate around LLM-based summarization and multi-speaker handling, since that's where the durable moat remains.

Related questions

More on Voice Memos

Deeper dives