Dictate
Dictate
Your voice, in writing. Nothing leaves your iPhone. Press once and talk. Dictate turns what you said into text on the device and cleans it up for reading.
iPhone and Apple Watch. Coming to the App Store.
- No cloud processing
- No account required
- No data collected
- No subscription fees
On your iPhone, nothing to set up
Parakeet, a leading open speech model, ships inside the app. It transcribes 25 languages and detects which one you speak, so you can switch languages without touching a setting. Apple Speech covers the other languages your iPhone has models for. Neither sends audio anywhere, and neither needs a key or an account. A fresh install records and transcribes before you open Settings.
A transcript you can read
Apple Intelligence makes a reading version on your iPhone:
- Filler words, stutters and false starts are removed.
- Spoken corrections are resolved: “Let’s meet on Tuesday, actually no, Wednesday at two” becomes “Let’s meet on Wednesday at 2:00 PM.”
- Counted points become numbered lists, lists of things become bullets, and dates, amounts and numbers get digits.
- Punctuation, casing and grammar are fixed.
The original is one tap away. It is split into timed blocks you can play from or copy one by one, and the words highlight as the audio plays. Add names and terms to your dictionary once, and the reading version spells them your way.
Meetings, with who said what
Mark a recording as a meeting while it runs, and Dictate labels the speakers on the device after you stop. The reading version keeps who said what, each speaker in their own colour.
One press, from anywhere
Start from your Apple Watch, the Action button, a shortcut, Control Center, the Home Screen widget or the Lock Screen. A Live Activity keeps the recording and its stop control on the Lock Screen while it runs. The Watch records on its own and hands the recording to your iPhone, now or as soon as the phone is back in reach.
Who uses Dictate?
Anyone with a lot on their mind
Reminders, ideas and to-dos, said out loud on a walk and waiting as a tidy list when you are back.
One pressJournalists and researchers
Interviews stay confidential because the audio never leaves the phone. No servers, nothing to leak.
100% on deviceProfessionals
Meetings and calls on speaker, with every speaker labelled on the device and a reading version to share.
Speaker labelsStudents
Lectures you can read back, play from any block, and search later in your history.
Timed blocksMultilingual speakers
Speak Spanish in the morning and German after lunch. Dictate detects the language each time.
25 languages, detectedBuilders
A private voice input for your own assistant, automation or backend, through a signed webhook.
WebhooksExtensible with webhooks
For people who want to take the text further: point Dictate at any HTTPS URL, and each original transcript arrives there as JSON, signed with HMAC-SHA256 so you can verify it came from your device. Failed deliveries retry on their own. Plain HTTP is accepted on localhost, .local hosts and private networks, so an agent on your own Wi-Fi works without a certificate. It is optional, and lives under Settings, Advanced.
Private by architecture
No accounts. No login. No servers of ours. No analytics. Transcription, speaker labels and the reading version all run on your iPhone. The only thing that can leave it is a transcript you send to your own webhook.
More
- Webhook reference: the payload, the signature, retries and a working receiver.
- Support: how it works, what to check first, and how to reach us.
- Privacy: what the app does with your data, in plain sentences.