Help

Tap the mic to start. Talk. Tap again to stop. Your words appear live while you speak; the polished version replaces them about a second after you stop. Talk as long as you like — one dictation holds up to 20 minutes.

Made a mistake mid-sentence? Just correct yourself out loud — “…friday, no wait, thursday” — and the cleanup applies it. Never re-start a dictation over a slip.

The first time, I ask before sending anything to my AI helpers: xAI turns your recording into words and Anthropic’s Claude polishes them, and neither keeps it. Change your answer anytime in Settings → AI helpers.

Hold Fn and talk — let go to finish. Double-tap for hands-free. Esc cancels. Your words land wherever your cursor is, in any app.

• ⇧⌘M works too: press it to start, press it again to finish.
• ⌃⌘V types your last dictation again; ⌃⌘C copies it.
• Drag the Murmur bar anywhere — it remembers where you put it.
• Change any of them in Settings → Shortcuts. On a Mac, if Fn opens emoji or Apple Dictation, set System Settings → Keyboard → “Press 🌐 key to” to “Do Nothing”.

Insert puts your words where your cursor was. On the Writing panel, Insert types them into the app you were using before Murmur, right where you left off. When Murmur knows that app, the button says so, like “Insert into Notes →”.

To do that, Murmur steps aside: its window hides so that app is in front again and your words land there. Murmur is still running. It hasn’t closed.

To bring Murmur back: Click Murmur in the menu bar and choose Open Murmur, or click Murmur in the Dock. Or press ⇧⌘M to dictate again right away.

Eight quick, hands-on lessons — about 9 minutes — that save you real time every day: thinking out loud, the right tone for each place, editing by voice, texting tools, teaching Murmur your words, snippets, turning a brain-dump into a to-do list, and using Murmur in every app. Each is about a minute, and you can skip, pause, or replay any of them.

Start the power-user training → (also in Settings → Power-user training)

Start with “Pip” and say what to change — “Pip, make that friendlier.” (“Murmur, …” works too.) No highlighting needed — I work out what you mean:

• Highlighted text? I edit just that.
• Say “the whole message”, “everything” or “all of it” and I edit all of it.
• Otherwise I edit the last thing you dictated.

“Pip” and “Murmur” only count at the very start — in the middle of a sentence they’re typed like any other word. “Pip” is a name too, so after it I listen for an instruction: “Pip, make that…”, “Pip, change…”, “Pip, remind me…” — while “Pip is coming to dinner” or “Hey Pip, how are you?” is typed as a message. “Scratch that” and “undo” work when they’re all you say; inside a longer sentence they’re typed (or, mid-thought, they fix what you just said). Every edit shows what changed, with an Undo.

Other languages: say “Pip,” with a little pause (or “Murmur”), then your instruction in any language. Taking things back works in Spanish, French, German and Portuguese too — “borra eso”, “efface ça”, “lösch das”, “apaga isso” (and “deshacer”, “annule”, “rückgängig”, “desfaz” to undo an edit).

Mac app: it works in any app after ⇧⌘M — I change the words I just typed, or whatever you’ve highlighted.

• Say what you mean. In Text or Email, say who it’s for and how it should sound — “Tell Brandon I looked at the numbers and I’m worried we’re overspending, but don’t make it sound harsh” — and I write the message itself, ready to send. I only do this when you clearly tell me who it’s for (“tell…”, “let … know”, “ask…”) or how it should sound (“make it kind”, “keep it short”), and I never add facts you didn’t say. Tap You said on the result to see your words next to what I wrote, or Use my words instead. In Notes and Formal, “tell Brandon…” stays as you said it. Turn it off in Settings → Style → Say what you mean.
• Tone buttons. Under every result: Warmer, Shorter, Firmer, Funnier — one tap, with the change shown and an Undo. Mac app: if I already typed it into another app, the button changes it there too.
• Reply from a screenshot. Tap Reply at the top (on a phone: Reply to a screenshot, under Try these), then choose, paste or drag in a screenshot of the conversation. Say or type the gist (“yes, but I’ll be late”) — or leave it blank for a few options — and I’ll write 2–3 replies in your voice to copy — or, in the Mac app, to type straight into the app you came from. Your screenshot is sent only to write the replies and never kept. Try it →On iPhone, start right in Messages: share a screenshot to Murmur. See the steps →

Tap Transcribe a file… (next to Transcribe) and choose a recording or a video — a voice memo, a meeting, a lecture. On a computer you can also drop the file anywhere on Pip’s screen. On iPhone and Android, share it to Murmur from Voice Memos, Files, Photos or any app.

• Up to an hour per file for now — for a longer one I offer to do the first hour. Most kinds work: .m4a, .mp3, .wav, .aac, .mp4, .mov, .ogg and more.
• You see how far along it is, with Cancel. If a part doesn’t come through, I say which part, and Retry sends just that part again.
• Then Copy, Polish (fillers and punctuation cleaned up, your words kept), Translate, Summarize, Read it back, or Export it as a .txt or Word file. It’s saved in History, named after the file.
• Or ask about it, or draft the follow-up email — see below.
• Your file stays on your device: its sound is sent in short pieces only to be written down, and nothing is kept. Privacy details →

Transcribe a file →

On a file’s transcript — or any long result, like a meeting you recorded with Transcribe — two more buttons:

• Ask about this: type a question, or tap the mic and say it — “What did we decide?”, “When is the budget due?”. I answer only from what was said, with short quotes from it. If it isn’t in the recording, I tell you. Ask follow-ups right there.
• Follow-up email: I draft the email you’d send after the meeting — a thank-you, what was decided, the next steps and who owns them — in your Email style, with your Dictionary’s spellings. If you made a summary, I write it from that. Edit it, try Warmer, Shorter or Firmer, then Copy or Open in Mail. I never add a name, date or promise that wasn’t said.
• Your questions, answers and the draft stay with the transcript in History on your device. Only its words are sent, to answer or write — never the sound — and nothing is kept. Privacy details →

Tap Transcribe and note a meeting and I write it down while it happens — who said what, and when — then make the notes: a summary, the key points and the action items.

• A call, in the Mac or Windows app: I hear the call through your computer’s sound (that’s “Others”) and you through your microphone (“You”). Headphones keep the two apart; without them, I leave out what your mic picks up from the speakers. On a Mac, choose Allow when macOS asks to let Murmur record your system audio.
• When a call starts — in Zoom, Teams, or a meeting in your browser — a small banner at the top of your screen asks if I should transcribe it. Press Start and I begin right away, with your microphone — and copy the heads-up line for you to paste in the meeting’s chat (unless you turned off Remind me to tell people). The arrow next to Start has Not this time and Don’t ask for that app; Esc closes it, and it goes by itself after 15 seconds. It’s kept out of screen shares and recordings where your computer allows it. I offer to transcribe — I never start by myself. Turn it off in Settings → Offer to transcribe a meeting when one starts (that’s also where the apps you said “Don’t ask” for are, to ask again).
• In person: on your phone or in any browser, tap Transcribe and note a meeting and I listen through the microphone — one transcript, with times (I don’t tell voices apart in a room). Keep Murmur open with the screen on.
• A meeting in another tab (Chrome or Edge on a computer, like Google Meet): choose “In another tab”, pick the meeting’s tab, and keep “Also share tab audio” on.
• Before you start, I remind you to tell everyone, with a line to copy and send. Some places require everyone’s permission to record — that part is up to you. Turn the reminder off in Settings → Remind me to tell people.
• While it records: a red dot, a timer, Pause and Stop, and the words appear as they’re written down. Up to 2 hours — I let you know at 1:50.
• After Stop: the notes and the transcript, with Ask about this, Follow-up email, Copy, Read it back and Export (.txt or Word). It’s saved in History as it goes — rename it with the pencil. If Murmur closes mid-meeting, what was written down is kept; the last minute or so may be lost.
• The sound is never kept: it’s sent in short pieces only to be written down. Privacy details →

Transcribe and note a meeting →

On a computer, tap Transcribe what’s playing (next to Transcribe a file…), press Start listening, then play the video, talk or podcast. I write down its words as it plays. Dictation can’t do this: it filters out sound from your own speakers so it hears only you.

• In the latest Mac or Windows app: I hear your computer’s own sound, and your microphone stays off. On a Mac, choose Allow when macOS asks to let Murmur record your system audio.
• In Chrome or Edge: choose the tab that’s playing and keep “Also share tab audio” on.
• Anywhere else, like Safari: I listen through the microphone, so play it out loud on your speakers, not in headphones, and keep the volume up.
• Then Copy, Summarize, Ask about this, Read it back or Export it. It’s saved in History as “What was playing”.
• Not on phones or tablets: they don’t let one app hear another app’s sound. Use Transcribe a file… with the recording instead.
• The sound is never kept: it’s sent in short pieces only to be written down. Privacy details →

Transcribe what’s playing →

Ready in the Murmur iPhone app. Tap one to see the steps.

• Dotted underline = not sure. When the words you saw while talking and Murmur’s final transcript disagree on a word, it gets a dotted underline. Tap it to pick the other word I heard, Type it… yourself, or say It’s right. Mac app: if I already typed it into another app, the fix lands there too. (No underlines in the Windows app or Firefox yet — they don’t show live words.)
• Fix it once. When you correct a name, product, acronym or other unusual word — by tapping an underline, with Fix, or by saying “Pip, change Viewgol to Vuegol” — I add the new spelling to your Dictionary, marked Learned, with an Undo. Everyday words are never learned. Fix the same two-to-four-word name the same way twice (“Atlyst Viewgol”) and I learn the whole thing.
• iPhone keyboard: words you fix while typing with the Murmur keyboard join your Dictionary too, next time you open the Murmur app.
• Your Dictionary: star the words you say most — starred and recently used words go to speech recognition first. Remove any word, learned or not, with its ×.
• Optional — help Murmur understand voices like yours. Off unless you turn it on in Settings. When it’s on, after you correct a dictation I ask whether to send that recording, with what I heard and your fix, to the Murmur team. Kept 12 months, never sold or shared, and you can delete them all from Settings anytime. Privacy details →

Style (Notes · Text · Email · Formal) shapes the result for its destination. Cleanup level sets how much editing happens: Off = untouched transcript, Light = punctuation only, Standard = fillers and corrections handled, Heavy = restructured into the target style. Check the Diff view on any result to see exactly what was changed — your raw words are always kept.

• The website added to your iPhone’s Home Screen (not the Murmur iPhone app): iOS sometimes asks for the mic again each time you open it — a known Safari quirk. If it gets annoying, use it in a Safari tab instead.
• A video playing on your computer comes out blank: dictation filters out your own speakers so it hears only you. Use Transcribe what’s playing instead.
• No live preview while speaking: Firefox doesn’t do live preview — your audio is still transcribed after you stop.
• Permission denied: re-enable the mic in your browser’s site settings (the padlock/ⓘ next to the address bar), or on iPhone under Settings → Apps → Safari → Microphone.
• Recording stops by itself: switching apps or locking the phone ends a dictation — keep the screen on while talking.

Something you wish Murmur did, or something that isn’t working? Suggest an idea or report a problem → Type it or dictate it — it goes straight to the Murmur team, and Kristen reads every one. Only what you write there is sent (plus which app and version you’re on), nothing else from Murmur.

Also on the main screen ( Suggest an idea) and in Settings.

Start the welcome tour any time to tell Murmur your name, who you text and your words — about 2 minutes, all by talking. (Also in Settings → Personalize Murmur.) Each device keeps its own, so on a new device you can skip it with “I already use Murmur” and bring your words over instead.

On the device you already use: Settings → Move my words to another device → Send from this device. Pick what to bring (your name, Dictionary, snippets, the people you text and your style are ticked; training, history and Scratchpad are up to you) and you get a code like 7K3M-QX9P-2HWT, plus a QR code.

On the new one: Settings → Bring my words from another device, and type the code. It works once, for 10 minutes, and adds to what’s there — nothing is removed. Your words are encrypted before they leave, so Murmur can’t read them. More

Getting set up, keyboard Full Access, Mac permissions, updates — Support has answers and a way to reach us. What happens to your words: Privacy.