Relay Support

Relay needs Microphone and Screen & System Audio Recording access. First-launch setup remains visible until both are enabled. You can review them in System Settings → Privacy & Security.

System audio is also used locally to reduce speaker playback picked up by the microphone. Only your selected audio sources are transcribed. Relay does not capture or save screen images.

Relay creates a folder in iCloud Drive → Relay for its .relay files. If iCloud Drive is unavailable, it uses a local folder shown as On My Mac. Click the folder location at the bottom of the sidebar to open it in Finder.

You can rename or move a file in Finder and reopen it in Relay. If you use iCloud Drive, Apple manages the file’s availability and synchronization.

Yes. Click Pause to stop capture and save the current transcript. Select the file in the sidebar and click Resume to continue, including after reopening Relay.

Only one Relay records at a time. If another is recording, Relay offers to pause it before switching. The timer counts captured time and excludes pauses and preparation.

Use Audio Sources or the language control in a conversation’s toolbar. Each file remembers its own choices. Changes made while recording save the current words and continue in that file with the updated choices.

On the new-conversation screen, your language choice becomes the starting language for the next new Relay. Changing a saved conversation’s language does not change that preference. Source and language changes apply to new speech; they do not change earlier transcription.

Auto Detect identifies the language from the opening speech, then uses that language for the conversation. It does not continuously switch between languages.

If the opening speech is too brief or unclear, choose a language directly. A language model may need to download before its first use; recognition and detection run on your Mac.

Selecting a passage copies the spoken words without source labels or timestamps. The selection keeps updating the clipboard while you drag, and a brief confirmation appears when you release the mouse.

After pausing, Copy Transcript and Export Transcript include the complete session with source labels and timestamps. Export is available as plain text or Markdown.

Microphone and System audio identify the audio input, rather than individual speakers. Speaker echo reduction helps reduce playback entering the microphone, including when only Microphone is selected.

An unfinished phrase is marked Live. Once recognition confirms its words and timing, Relay displays its timestamp and places it chronologically.

Swipe left on a transcript in the sidebar, or use its context menu, to reveal Rename and Delete. Deletion moves the file to the Trash. Deleting the recording Relay stops its capture. Deleting a different Relay leaves the current recording running.

Relay files can also be managed in Finder. Review the Trash before emptying it if you may want to restore a deleted file.

Relay requires macOS 26 or later and a Mac that supports Apple’s on-device speech recognition. Available languages and language models depend on your macOS version.

If Relay reports that recognition is unavailable, contact support with your Mac model and macOS version.

Speech recognition and language detection happen on your Mac. Audio is held in memory during processing and is not saved as a recording. Relay does not send your audio or transcripts to Asher Bloom.

Language models may download before use, and transcript files saved in iCloud Drive are handled by Apple. Read the Privacy Policy for details.