Recording & Live Captions

Subwright records from your microphone or macOS system audio and shows live transcription while you record. The floating caption window can also display original and translated text side by side.

Recording sources

Microphone: Record your voice on Windows, macOS, or Linux
System Audio: Record what your computer is playing; currently macOS only (requires Screen Recording permission)

Windows and Linux do not currently capture system audio. The recording page says so explicitly and will not start a system-audio-only session.

Start recording

Open Record (/record) from the sidebar
Choose an input device:
- Microphone: select your microphone
- System Audio: select system audio capture (macOS only, requires authorization)
Click “Start Recording”
Watch the live waveform and incremental text while recording
Click “Stop Recording” — the final tail is finalized too

Live transcription

See real-time transcription while recording:

Transcribe as you record, no waiting for the file to finish
Great for meetings, interviews, and live streams
Live failures are no longer silent; a dropped model shows a notice and resumes automatically after recovery

Floating bilingual captions

Open the floating caption window to:

Keep the original text on top and follow the player or system audio
Turn on “Live translation” to stream the translation under each line
See a clear status for translating, failure, or missing AI endpoint
Auto-translate after stopping and save the final transcript to history

Live translation requires a configured LLM endpoint in Settings → AI.

Dual-source recording

On macOS you can select the microphone and system audio together:

Both sources are mixed before transcription
Speakers can be picked up by the microphone; headphones are recommended
Echo cancellation is not claimed; the UI says so honestly

Permissions

macOS

Microphone: System will prompt on first use — click “Allow”
System Audio: macOS requires “Screen Recording” permission. Subwright guides you to System Settings if not authorized

Windows / Linux

Microphone: Windows or your desktop environment requests permission on first use
System Audio: Not supported yet; the recording page explains this

Recording settings

In Settings → Recording you can adjust:

Default recording device
Recording save path
Auto-transcribe toggle
Live translation and floating caption behavior