Earshot for iPhone and iPad — Privacy Policy

Last updated: September 27, 2026

Earshot (listed as “Say it now” in TestFlight) is developed by Al Chang. This policy covers the iPhone and iPad app and its keyboard extension. For questions or privacy requests, contact al@potdot.com.

Dictation and microphone access

Earshot uses the microphone with your permission to capture speech you choose to dictate. The main app captures audio because iOS keyboard extensions cannot access the microphone directly.

After a dictation, the microphone may remain active for up to ten minutes of idle time so you can dictate again while using another app. iOS displays its microphone indicator while access is active. Between dictations, Earshot processes incoming microphone buffers, including audio levels, but does not add that audio to a dictation recording or submit it for transcription.

To stop microphone access immediately while idle, open Earshot Settings → Voice model → Stop microphone. You can also revoke microphone permission in iOS Settings.

On-device transcription

Parakeet is the default transcription option. After you download the voice model, it processes dictation on your device without sending the recording to a transcription service or to the developer.

Installing the model downloads files from Hugging Face and its delivery infrastructure. These providers receive ordinary network information, such as your IP address and the requested files. Your dictation audio is not part of this download. See Hugging Face’s privacy policy.

Audio is held temporarily for transcription. On-device transcription also uses a temporary audio file that Earshot attempts to remove when processing finishes or returns an error. An interrupted process may leave temporary files until they are removed by the app or operating system. Earshot does not maintain a recording library.

Optional xAI cloud transcription

If you select Grok and supply your own xAI API key, Earshot sends the dictation recording and selected transcription settings directly to xAI over HTTPS. xAI receives the request, including your API credentials, network information, and audio, and returns a transcript. Verifying an API key also contacts xAI. These requests do not pass through a developer-operated server.

xAI’s handling and retention of this data depend on its terms and your account settings. Earshot does not promise zero retention by xAI. Review xAI’s privacy policy and API data-handling information before choosing this option. Select Parakeet to keep future transcription on-device.

Information stored on your device

Earshot stores transcript text, timestamps, duration, language, and preferences locally. The app and keyboard share settings, recording status, audio-level information, and pending transcript text so the keyboard can insert your dictation into the selected text field. Full Access enables communication needed by the dictation feature; it does not cause Earshot to upload everything you type to the developer.

Your optional xAI API key is stored in the iOS Keychain. When you insert or copy a transcript into another app, that app’s handling of the text is governed by its own policies. Device backups may include app data according to your Apple settings; Earshot does not provide its own cloud history-sync service.

Diagnostics and beta feedback

Earshot writes local diagnostic logs about operation, errors, and performance. Release builds redact transcript text at the app’s designated content-logging points. Diagnostics or screenshots you choose to share may contain personal information, so review them before sending them.

When you use TestFlight, Apple collects and shares beta-testing information with the developer under Apple’s TestFlight privacy terms. If you email feedback, the developer receives your email address, message, and any attachments and uses them to respond and troubleshoot. Support correspondence is retained as needed to address the request and maintain relevant support records; contact the developer to request deletion.

Earshot does not include advertising or developer-operated behavioral analytics. The developer does not sell your personal information or use your dictation to train AI models.

Your controls and deletion

Delete a saved transcript by opening its context menu in Earshot history and selecting Delete. A transcript waiting for insertion is a separate temporary copy shared with the keyboard; deleting a history entry does not necessarily clear that copy or text already inserted into another app.

You can delete the downloaded voice model in Earshot Settings, remove the keyboard in iOS Settings, and revoke microphone permission. Revoke your xAI API key in your xAI account to disable its use. Keychain items and device backups may persist independently of deleting the app. Deleting local data does not delete copies already sent to xAI, inserted into another app, or included in feedback; use the relevant provider’s controls for those copies.

This policy website

This page is hosted by Cloudflare. When you visit it, Cloudflare processes request information such as your IP address and browser details to deliver and protect the site. The page does not add analytics, advertising, or cookies. See Cloudflare’s privacy policy.

Changes

This page will be updated when Earshot’s data handling changes, with the revision date shown above.