An offline transcription app that still knows who said what

Download one speech model, then transcribe with the network switched off — on a plane, in a basement meeting room, in a field interview two hours from the nearest tower. Speaker labels, translation and search all work exactly the same.

The Murmur library with a recording in progress

Offline is a consequence here, not a mode

Plenty of apps advertise an offline mode that only queues your recording until the signal comes back; the transcription itself still happens on a server later, and until it does you have an audio file and a spinner. Murmur has nothing to queue. The speech recognition model is downloaded to the device once and runs there from then on, so a recording made in airplane mode is already a finished, searchable, speaker-labeled transcript before you land. The network is used to fetch a model in the first place and to update the app — that is the whole list.

One download, then nothing
Pick a speech model once over Wi-Fi. After that, transcription, speaker separation and translation all run locally with the radios off.
No queue and no processing screen
The transcript appears as the phone works through the audio. There is no server-side job to wait on and nowhere for your recording to get stuck.
Planes, basements, fieldwork
The situations where a recording matters most are often the ones with the worst connectivity. Those are exactly the ones an on-device model is indifferent to.
Long sessions, without filling the phone
Recordings keep running in the background, and the audio is compressed automatically after transcription so a term of lectures does not eat your storage.

Side by side

On-device app vs cloud transcription service

Murmur compared with a typical cloud transcription service
CapabilityMurmur (on-device)Typical cloud service
Where transcription happensOn your own deviceOn the provider’s servers
Your recording is uploadedNoYes — that is how it works
Account requiredNoUsually
Works with no signalYes, after a one-time model downloadNo
Speaker labelsOn-device, remembered across recordingsServer-side, if the plan includes it
Transcript translationOn-device, line by lineServer-side, where offered
Transcribe a file you already haveYes, on-deviceYes, by uploading it
Export formatsPlain text, Markdown, PDF, WordVaries by plan
Who else can read the transcriptNobody unless you share itYou and the provider
If the service is down or shuts downNothing changes — it all runs locallyYou lose access

What still needs a connection

Two things, both optional and both one-off. Speech models are downloaded from inside the app rather than shipped in it, so the first run needs a network — after that the model is on your device and stays there. And if you turn on iCloud sync it obviously talks to iCloud: your own private iCloud account, syncing your own files between your own devices. Recording, transcription, speaker identification, translation, search and export never touch the network at all.

Offline usually means fewer features. Not here.

The normal trade for working offline is a stripped-down version of the product. An offline transcript in Murmur still carries timestamps, still labels each line with the person who said it, still recognizes people from earlier recordings, still searches, and can still be translated line by line and read next to the original. Nothing degrades when the signal goes, because none of it was reaching for the network in the first place.

What you need to run it

Murmur runs on iPhone, iPad and Android — iOS 18 or later on Apple devices, Android 8.0 or later. A speech model takes up space on the device, so the practical constraint is the storage you are willing to give a recorder, not the connection you happen to have that day. The app is free to download, and none of the on-device work needs an account.

FAQ

Offline transcription app — common questions

Does it really transcribe in airplane mode?
Yes, once a speech model has been downloaded. Turn the radios off and record — the transcript is produced on the device while you are recording it.
How much has to be downloaded first?
One speech model, chosen inside the app. It is a one-time download, and after that nothing else has to come over the network for transcription to work.
Do speaker labels work offline?
Yes. Voice separation and the library of people you have named both live on the device, so they behave identically with no signal.
Can I translate a transcript offline?
Yes. Transcript translation runs on-device as well, so it works with the network off.

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