Does Apple dictation send your audio to Apple?
Apple's dictation sends your audio to Apple's servers or processes it entirely on your device, depending on your device, language, and a setting you can check in Keyboard Settings or Siri Settings. Apple's own privacy pages describe both paths. Parlo, by contrast, always transcribes and cleans up speech on your device, with no cloud step.
Whether Apple's dictation sends your audio to Apple depends on your device, your language, and a setting, not on a single fixed answer. This guide reads Apple's own published pages to lay out what they say, then explains how Parlo's approach differs.
Does Apple dictation ever send audio to Apple's servers?
Yes, depending on the device, language, and setting. Apple's privacy page for Ask Siri and Dictation states that "your device will indicate in Keyboard Settings if your audio and transcripts are processed on your device and not sent to Apple servers." The same page adds: "Otherwise, the things you dictate are sent to and processed on the server, but will not be stored unless you opt in to Improve Siri and Dictation." The Mac dictation guide describes the same setting for Mac, telling users to check "whether your voice inputs and transcripts for general text Dictation ... are processed on your device and not sent to Siri servers." Both pages describe on-device and server processing as two paths Dictation can take, depending on your setup, rather than one fixed behavior for every device.
How do you check which path your device uses?
Apple's pages point you to Keyboard Settings or Siri Settings for the on-device indicator, without spelling out an exact menu path that holds across every iOS and macOS version. The practical way to check is to look in your device's Siri or Keyboard settings for language along the lines of "on-device" or "not sent to Apple servers." Apple describes that indicator as the source of truth rather than a fixed rule you can look up once for every device.
What does Apple do with audio it collects?
Apple's Improve Siri and Dictation privacy page describes an optional setting, separate from Dictation working at all, that if turned on retains "audio data, transcripts, and related request data, such as the category of request you made, device specifications, performance statistics" for up to two years, longer for a reviewed subset: "the small subset of interactions that has been reviewed may be kept beyond two years." The same page states this data "is associated with a random, device-generated identifier that rotates multiple times per hour and is not tied to your Apple Account or email address."
How is Parlo different from Apple's dictation?
| Category | Apple dictation | Parlo |
|---|---|---|
| Where processing happens | On-device or sent to Apple's servers, depending on device, language, and a setting | On-device, every time |
| Where you check the setting | Keyboard Settings or Siri Settings, per Apple's own pages | No setting to check; there is no cloud path in Parlo's pipeline |
| Optional data collection | Improve Siri and Dictation, retained up to two years if turned on, longer for a reviewed subset | None. Audio is discarded right after transcription |
| Cleanup step | Not described on Apple's pages | An on-device language model removes filler sounds and false starts |
Parlo transcribes with Apple's own SpeechAnalyzer and SpeechTranscriber, the same on-device engine family behind the newest system dictation, and runs it on your device every time, not as one of two possible paths. Cleanup then runs on-device too, with Apple's FoundationModels framework.
Does turning on history sync change where your audio goes?
No. History sync is about your finished transcript, not your audio, which is discarded before sync ever enters the picture. If sync is on, Parlo encrypts the transcript on your device first, using a key kept in iCloud Keychain, then stores the ciphertext in your private CloudKit database. Neither Apple nor Parlo can read that ciphertext. Sync is on by default and end-to-end encrypted; you can turn it off in Settings, or by not signing in with Apple.
Should you avoid Apple's dictation because of this?
Not necessarily. Apple documents Dictation on iPhone, iPad, and Mac, and the same on-device-or-server setting appears in Apple's privacy pages for all three. Apple's own pages describe an on-device path that exists for many configurations, and using Dictation at all is a personal call based on what its published pages tell you. This guide exists because that answer depends on settings most people never open, not because Apple's dictation is unsafe to use. If you want processing that is on-device without checking a setting first, Parlo's pipeline works the same way every time it runs, on iPhone and on Mac. You can try it from the beta page.
Does the language you dictate in change anything?
Apple's Mac guide states plainly that "Dictation is not available in all languages or regions, and features may vary," without listing which languages use which processing path. Parlo does not vary by language either, though for a different reason. It is English-first today, with transcription falling back to en-US when a requested locale is not available, and there is no language picker yet.
What about an older Apple support article that mentions server access?
A separate, older Apple support article, checked the same day as the pages above, states that Dictation "needs to access certain Apple servers using your Internet connection" unless a legacy "Enhanced Dictation" offline option is turned on first. That wording is from an earlier era of macOS and predates the newer on-device Siri and Dictation setting described on Apple's current privacy pages. It is included here for completeness, not because it overrides the newer pages.
If you find an Apple page that states something different from what is quoted above, checked after 2026-09-06, reply to beta@parlokey.com and the team will update this guide.
Frequently asked questions
Where does Apple say whether Dictation is on-device or sent to a server?
Apple's privacy pages for Siri and Dictation say your device indicates in Settings whether audio and transcripts are processed on-device or sent to Apple's servers.
Does Apple ever store the audio it collects?
Apple's privacy page describes an optional Improve Siri and Dictation setting, which if turned on retains audio, transcripts, and related request data for up to two years, longer for a reviewed subset.
Is the data Apple collects tied to your Apple Account?
No. Apple's privacy page says this data is associated with a random, device-generated identifier that rotates multiple times per hour, and is not tied to your Apple Account or email address.
Does Parlo ever send your dictation audio anywhere?
No. Parlo captures audio, transcribes and cleans it up on your device, and discards the recording. There is no cloud step in that pipeline.
Does Parlo's cleanup model run on a server?
No. Cleanup runs on-device with Apple's FoundationModels framework.
Can you turn off Apple's data collection for Dictation?
Improve Siri and Dictation is a setting you can turn off. Apple's pages describe it as separate from Dictation working at all.