Recording Indicator
Whenever Murmur is listening, an indicator appears on screen so you always know when the microphone is live, what has been heard so far, and where the text will land.
Indicator styles
Choose a style in Settings > Dictation (and separately for voice notes in Settings > Notes):
- Notch — extends downward from the MacBook notch area. The default; feels native on notched MacBooks and stays out of your content.
- Bottom Bar — a wide bar near the bottom of the screen. The most readable style if you want to watch the streaming transcription as you speak.
- Compact — a small bar in the top-right corner. The most minimal option.
What the indicator shows
- Live waveform with an audio level meter, confirming the microphone is picking you up
- Streaming partial text as you speak, right-aligned so the newest words appear at the right edge (engines that support streaming only — see Engine Selection)
- The target app that will receive the text
The indicator appears instantly the moment you trigger dictation — before the model finishes warming up — so a tap always gets immediate feedback. If the model is already loaded, it skips straight to recording rather than flashing "Preparing."
Shows on every screen
(New in 1.6) If you use more than one display, the indicator now appears on all of them at once, so it is always in view no matter which screen you're working on. It repositions itself automatically when you change your display arrangement or resolution.
Hover controls
(New in 1.6) Move your pointer over the indicator and two controls fade in:
- Stop & finish (green) — stop recording and process the transcription normally.
- Cancel (red) — discard the recording.
These give you a mouse-driven way to stop or cancel that works independently of the keyboard, so a session can always be ended even if a hotkey misbehaves. Which controls appear depends on the stage: while recording you get both; once Murmur is uploading, processing, or refining, only Cancel remains.
Status states
The indicator walks through the lifecycle of every dictation:
| State | Meaning |
|---|---|
| Preparing | The transcription model is loading into memory — happens on first use after launch, takes a few seconds |
| Recording / Listening | The microphone is live |
| Reacquiring audio | (New in 1.6) Your microphone changed — Murmur is switching to the new device and will resume in a moment. See Automatic Device Switching |
| Processing | Audio is being transcribed |
| Refining | The transcription is being polished by your AI workflow |
| Pasted / Copied | Done — text was pasted into the target app, or copied to the clipboard |
| Cancelled | You cancelled the recording |
| Error | Something failed; check Common Issues |
In Auto mode the indicator also coaches you through the press: it first shows "Release ⌥" while you hold, then "Tap ⌥" once a quick tap has locked in toggle mode.
Gestures while the indicator is up
Escape behaves differently depending on what you are doing:
| Gesture | Dictating | Command mode | Raw mode |
|---|---|---|---|
| Single Escape | Step down: stop and paste the raw transcription, skipping refinement | Cancel | Cancel |
| Double Escape (within 0.5 s) | Cancel | Cancel | Cancel |
| Escape during model warmup | Cancel | Cancel | Cancel |
| Escape while holding a push-to-talk key | Cancel | Cancel | Cancel |
Shift, tapped bare (not as part of a Shift+key combination, which is ignored):
- While dictating — switch to Command Mode
- In command mode — switch back to Dictation
- In raw mode — return to the mode you were in before
Cancel does not destroy your recording. A cancelled session above a minimum length is saved to History, marked cancelled, with its audio intact — you can re-transcribe it later. Note that while holding a push-to-talk key, Escape always cancels outright; the step-down-to-raw gesture is available in toggle and auto-toggle recordings.
The indicator never captures or displays content from other apps — it only shows your own audio levels, your own words, and the name of the app you are dictating into.