Offline Speech to Text

Whisper Notes is an offline speech to text app for iPhone and Mac. The speech model runs on your device, so it works without an internet connection and your audio is never uploaded. Record from the Lock Screen, import an audio file, or transcribe a meeting; the text appears on the same device.

What works offline?

Everything. The default Parakeet model is built into the app, so it works offline from the first launch. Recording, transcription, speaker labels, search and export all run on the device, so airplane mode works. We never see your recordings or transcripts, and they're never used to train AI.

On iPhone, one tap on the Lock Screen widget opens the app and starts recording, and a Live Activity shows the recording on the Lock Screen and in the Dynamic Island. You can also import audio files, or share a recording from Voice Memos into Whisper Notes.

On Mac, you can record, import audio and video files, and record Zoom, Teams or Google Meet calls from system audio. In the Mac app downloaded from our website, you can also hold Fn to type with your voice in any app.

Add names, product terms and jargon as custom vocabulary in Settings, and the model spells them the way you do. To move text out, copy it, share it, or select several transcripts and export them together.

Which speech model runs where?

On iPhone, Whisper Notes ships three speech engines, plus Apple Speech on iOS 26; on Mac it also has Qwen3-ASR 1.7B (Beta), and Apple Speech on macOS 26 or later. You choose one in Settings; an engine you add downloads once.

•Parakeet is the default: the fastest engine for English and 24 other European languages, built into the app. It is Moondream's Parakeet Ultra, post-trained from NVIDIA Parakeet V3. On English, Parakeet V3 scored 6.32% WER on the Open ASR Leaderboard and ran about 60× realtime on an M5 MacBook Air.
•Whisper covers 100+ languages. The Mac runs Whisper Large V3 Turbo (about 2 GB); the iPhone runs Whisper Small, which has the same language list.
•SenseVoice is the fast option for Chinese, Japanese, Korean and Cantonese. On an M4 Pro Mac it transcribed a 27-minute Chinese podcast in 13.83 seconds (118×). That test ran on an earlier version of the app; today's model card speed, 115–131× realtime, is about the same. For the fewest mistakes in those languages, the Mac also has Qwen3-ASR 1.7B (Beta).

On a Mac, the built-in Parakeet runs at about 74× real time: an hour of audio in under a minute. On iPhone, the App Store listing reports about 18 seconds for five minutes of audio with Parakeet on an iPhone 15. The full list of languages each engine supports is on one page.

What does it cost?

On iPhone, Whisper Notes is $7.99 once. On Mac, the first 5,000 words every week are free, then it is $14 once. Neither has a subscription or an account, and every engine is included.

What are the limits?

•Apple devices only. iPhone 12 or later with iOS 18, or a Mac with macOS 14, built for Apple silicon (the Mac App Store version also installs on Intel Macs). There is no Android or Windows version.
•Speed depends on the device. An hour-long recording takes minutes on an iPhone, not seconds, and older iPhones are slower. If you need a long file back instantly, a cloud service is still faster.
•Keep the iPhone app open. With Whisper or SenseVoice, transcription stops when you switch to another app; Parakeet and Apple Speech can usually finish a short job within about 30 seconds in the background, if iOS allows it. Otherwise, come back to Whisper Notes and tap the retry icon on the note. A recording you stop from the Lock Screen, Apple Watch or Shortcuts is transcribed the next time you open Whisper Notes.
•No sync. Recordings stay on the device where you made them. Use export or AirDrop to move them.

If one of these rules Whisper Notes out, our guide to the best transcription apps for Mac and iPhone compares the other offline and cloud apps.

Offline Speech to Text: FAQ

Does speech to text work without internet?

Yes. The default Parakeet model is built into the app; Whisper (Large V3 Turbo, ~2 GB, on Mac; Small on iPhone) and SenseVoice are optional downloads that you fetch once. After that, recording and transcription run entirely on-device: airplane mode works, and zero data is used.