Private because it is local
No account, no sign-up, no upload. The speech model runs on your device, so your recordings and transcripts stay yours by architecture, not by policy.
100% on-device · No account · Works offline
The voice recorder that transcribes offline — and knows who said what.
Murmur turns meetings, interviews and lectures into searchable text with every speaker labeled, entirely on your own iPhone, iPad or Android device. Nothing is uploaded. Nothing needs a signal.
- Free to download
- No sign-up
- 100+ languages
- Export to PDF, Word, Markdown
100+spoken languages transcribed
0recordings sent to our servers
4export formats, speakers included
3devices: iPhone, iPad, Android
Why Murmur
A recorder should hand you the words, not another audio file you will never replay.
Most transcription apps ship your conversation to somebody else's server to do it. Murmur does the whole job — listening, writing, separating voices — on the phone in your hand.
Knows who said what
Each line carries the name of the person who said it, color-coded. Murmur builds a library of voices, so people it has met before are recognized next time.
Works with no signal
A plane, a basement meeting room, a field interview in the middle of nowhere. Download a model once and Murmur keeps transcribing.
Features
Everything you need between “hit record” and “here are the notes”.
Record
Hit record, get the words
A familiar voice recorder that happens to write everything down. Start a recording and Murmur transcribes it on-device, detecting the spoken language automatically.
- Long recordings keep running in the background
- Automatic language detection
- Audio is compressed after transcription to reclaim storage
Speakers
It remembers people across recordings
Name a voice once and Murmur recognizes it the next time that person speaks. Merge duplicates, relearn a voice, and the identification keeps getting better.
- A people library shared across every recording
- Rename, merge or relearn any speaker
- Color-coded labels on every transcript line
Translate
Read the conversation in your language
Turn any transcript into another language line by line, on-device. Read translation-only, or side by side with the original for study and review.
- Line-by-line, keeps the speaker labels
- Bilingual or translation-only view
Export
Export to PDF, Word, Markdown — or hand it to your AI
Take the transcript wherever the work happens: a document, an email, your notes app, or an AI assistant that summarizes it and pulls out action items. Speaker names and translations come along.
- Plain text · Markdown · PDF · Word
- Share the original audio too
- Your tools, your choice of AI — nothing is forced through us
Search
Every word you ever recorded, findable
Search inside a transcript or across the library and jump straight to the moment. Matches are highlighted exactly where they appear.
- Full-text search with highlighted hits
- Import existing audio and video files to transcribe
Playback & corrections
Replay a sentence, or fix who said it
The transcript scrolls along with the audio as it plays, and any single line can be replayed on its own. If a line was attributed to the wrong person, reassign it — for that line, or for every line by that speaker.
- Replay one sentence at a time
- Transcript follows playback, at the speed you choose
- Reassign a line to another speaker, or fix several at once
Privacy
Your conversations never leave your device.
This is not a promise about how we handle your data. It is a statement about where the work happens: on your phone, by a model on your phone, with no server in the loop.
- 100% on-device transcription and speaker identification
- No account and no sign-up
- Your recordings are never sent to our servers
- Optional iCloud sync uses your private iCloud account — we never see it
- Works fully offline once a model is downloaded
Read the full privacy policy.
How it works
Three steps, no setup ritual.
- 01
Hit record
Open Murmur and record the meeting, the interview, the lecture, or a thought on a walk. Long takes keep running in the background.
- 02
Murmur writes it down
The speech model runs on your device, detects the spoken language and produces a timestamped transcript with each speaker labeled.
- 03
Read, search, export
Tap a line to replay it, search the whole transcript, translate it, then export to PDF, Word, Markdown or plain text — or hand it to your AI assistant.
Use cases
Built for the conversations you cannot afford to half-remember.
Meeting transcription
Stay in the conversation instead of taking notes. Every decision is in the transcript, attributed to the person who made it.
See meeting transcriptionInterview transcription
Journalists and researchers get a verbatim, speaker-separated record that never leaves the device — sources stay confidential.
See interview transcriptionLecture notes
Record the lecture, read it back with the audio scrolling in sync, and search for the one definition you need before the exam.
See lecture notesLanguage learning
Import a speech or a podcast, tap any sentence to replay it, and read the translation line by line next to the original.
See language learningAudio file transcription
Already have the recording? Import a voice memo, a meeting file or a video and read it back as text, speakers included.
See audio file transcriptionLanguages
100+ spoken languages, detected for you.
Record in whatever language the room is speaking — Murmur works out which one it is. The app interface itself ships in English, Japanese, Simplified Chinese, Korean and French.
- English
- 日本語
- 简体中文
- Español
- Français
- Deutsch
- 한국어
- Português
- Italiano
- Русский
- العربية
- हिन्दी
- Nederlands
- Polski
- Türkçe
- Tiếng Việt
- ไทย
- Bahasa Indonesia
- Svenska
- Українська
- and ~90 more
FAQ
Questions people ask before downloading.
Are my recordings uploaded anywhere?
No. Recording, transcription, speaker identification and translation all run on your device. Your recordings and transcripts are never sent to our servers.
Does it work offline?
Yes — that is the point of running locally. Once the model is downloaded, Murmur transcribes on a plane, in a basement, or anywhere with no signal.
How does Murmur know who is speaking?
It separates the distinct voices in a recording and labels every transcript line with the speaker who said it, color-coded so a multi-person conversation is easy to follow.
Which languages can it transcribe?
100+ spoken languages, detected automatically from the audio. The app interface is available in English, Japanese, Simplified Chinese, Korean and French.
Can I transcribe a file I already have?
Yes. Import any audio or video file — a voice memo, a meeting recording, a lecture — and Murmur transcribes it on-device with speakers labeled.
What formats can I export?
Plain text, Markdown, PDF, Word — each line carrying its speaker, plus your translation if you made one. You can share the original audio file too.
More answers on the full FAQ page.
Start with your next conversation.
Download Murmur, download a speech model once, and every recording after that is transcribed on your own device — meetings, interviews, lectures, ideas on a walk.
Free to download · No account · Works with no signal