A little less typing.

Make yourself heard.

Start with your original words

1. Open Sayso and allow microphone access when prompted.
2. Apple Speech is selected by default. Choose a supported language in Settings and allow any required speech assets to finish downloading.
3. Tap the recording button, speak, then tap Stop. Original keeps the transcript. Choose Clean, Message, Email, Notes or a custom writing mode to rewrite it with Apple Intelligence.
4. Review and edit the result, then use Copy or Share. Open History to find saved dictations.

Use the optional Parakeet speech model

In Settings, select Parakeet · local under Speech model. Tap Download Parakeet (about 500 MB), or import its Core ML model folder from Files. The download comes from Hugging Face and can take a few minutes. After setup, Parakeet transcribes locally after you tap Stop and supports recordings up to ten minutes. It automatically detects 25 European languages, including English and Dutch; it does not support Arabic, Chinese, Japanese or Korean.

No words appearing?

Check microphone access in iPhone Settings, confirm that your selected speech model and language are ready, and try the built-in microphone with Bluetooth disconnected. With Parakeet, wait until after Stop for the transcript. If the issue continues, open a GitHub issue with your iPhone model, iOS version, app version, and the error shown. Do not include private recordings or transcripts.

Rewriting is unavailable

Rewriting needs Apple Intelligence enabled on a compatible iPhone, with its on-device model ready and a supported language. Availability can vary with Apple's device, language and region support. Original remains available when your selected speech model is ready. Sayso requires iOS 27 or later. Review rewritten text before using it; your original transcript remains available.

Make a writing mode your own

Open Modes and choose Edit prompt or Add mode, or go to Settings → Writing modes. You can adjust built-in prompts, restore them, or add a named mode with your own instructions. Vocabulary in Settings gives Apple Speech and rewriting extra context for names and terms.

Start from Shortcuts or the Action button

Add Sayso's Start Dictation action in Shortcuts. You can assign that shortcut to the Action button. The action opens Sayso to record. Recording finishes when Sayso moves to the background.

Get in touch

Sayso is developed by Ali Soliman. For help or privacy questions, open a GitHub issue. Issues are public; do not include sensitive information.

Your words stay yours.

Privacy policy · Updated 20 September 2026

Sayso uses Apple's on-device speech technology or the optional local Parakeet model to turn speech into text, and optionally Apple's on-device Foundation Models framework to rewrite it. No Sayso account or API key is required. Sayso does not operate a server to receive your audio, transcripts, vocabulary, or rewritten text, and includes no advertising or third-party analytics.

Audio and processing

Microphone access provides live recording and transcription. Sayso processes microphone audio in memory and does not save microphone recordings. Transcription and rewriting run on your device.

Model downloads and external websites

Apple speech and Apple Intelligence assets may need an initial internet download through Apple services. If you explicitly choose Download Parakeet, FluidAudio downloads the model files from Fluid Inference's repository on Hugging Face over HTTPS. Your audio, transcripts and vocabulary are not sent with this download. The download host and its delivery providers receive the network information needed to deliver the files, such as your IP address and the requested files, under their own privacy practices. You can instead import an existing model folder and remove the installed Parakeet model in Settings.

Opening a model attribution or support link visits an external website. See Hugging Face's privacy policy and GitHub's privacy statement.

History and preferences

If you enable history, dictations are stored locally and may be included in your device backup. You can delete saved dictations in the app. Turning history off prevents new entries from being saved; existing entries remain until deleted. Preferences, vocabulary and writing modes are also stored locally.

Copying and sharing

Copy places your chosen text on the local-device pasteboard. Share discloses text to your chosen destination, which handles it under its own privacy practices.

Diagnostics

Copy diagnostics explicitly copies technical metadata and recording counters to the local-device pasteboard. It does not include audio, transcripts, vocabulary, input-device names, or route identifiers. Sayso does not automatically send diagnostics to a server.

Support correspondence

Support is handled through GitHub Issues. Your GitHub username and anything you post in an issue are public and handled under GitHub's privacy practices. Do not post recordings, transcripts, contact details, or other sensitive information. You can edit or delete your own comments using GitHub's controls.

Questions or requests

For questions about this policy, open a GitHub issue without sharing private information.

Third-party notices

See the open-source and model acknowledgments for attribution, source links, and copies of the applicable license and notice texts.