How to transcribe a voice memo on an iPhone?
Open the memo in an app that transcribes it, or record inside that app to begin with. Voice Memos saves each recording as an .m4a file on the phone, so handing it to another app is a share rather than an upload. With Kaseta the recognition runs on the iPhone itself once its models are downloaded, so the memo is never sent anywhere and the transcript arrives whether or not you have signal.
What is the best transcription app for iPhone?
It depends on where you want the audio to end up. Otter and Notta are the stronger choice if you need a shared team workspace, transcripts colleagues can open in a browser, and a bot that joins your calls and hears every participant at source, and they charge a subscription for it. Kaseta is the better fit when the recording should not leave your phone, or when you would rather not rent access to your own notes by the month.
Can I transcribe audio offline?
Yes, once the recognition models are on the phone. Kaseta downloads them on first launch – that is the one step that needs the internet – and after that airplane mode, a plane and a basement seminar room all behave the same as full Wi-Fi. The check takes a minute and is worth doing yourself: switch every radio off, record something, tap Stop, and wait for the text.
Can you convert M4A to text?
Not by converting the file. A format converter rewrites the container and the result is still audio. Getting text means running speech recognition over the sound, which is a different operation: either an app on your device does it, or a service does it on a server after you upload the file. On an iPhone nothing needs converting first, because .m4a is already what Voice Memos saves.
What is the best free voice recorder with transcription?
Apple's Voice Memos is free, already installed, and transcribes short notes in the languages Apple covers. Kaseta's free plan records with no limit at all and transcribes the first five minutes of each recording. The ceiling is on the text and never on the recording, which keeps going to the end and is saved in full.
Does the app send my recordings anywhere?
No. Your audio is never uploaded for transcription – the recognition runs on the phone. The app does go online for two things that carry no audio: downloading the recognition models, and App Store purchases through Apple. One thing worth knowing is not about the app at all: if iPhone backup is switched on, your recordings are included in your own iCloud backup like everything else on the phone. That copy goes to your Apple account, never to us, and you can exclude Kaseta in the iPhone's iCloud settings.
Do I need an account?
No. No sign-up, no email address and no login screen, because an account would imply a server that knows who you are. That also means there is no account for anyone to subpoena, and no password reset that could hand your recordings to someone else. This website sets no cookies either.
How many languages does it transcribe, and is the app in my language?
Two different numbers, and it is worth not merging them. The transcription engine transcribes 99 languages, every one of them recognised on the phone. The app's own interface ships in 39 App Store localisations, which are 33 languages once regional variants are counted once – a much shorter list either way. Transcription in your language works whatever language the buttons happen to be written in.
Is there a limit on how long I can record?
No, on any plan. Recording and import are never cut off – that is a fixed property of the app rather than a plan feature. The text is what has a ceiling on the free plan: the first five minutes of each recording. Pro transcribes the whole recording. A two-hour lecture on the free plan still records to the end and transcribes its opening stretch.
Can I export the transcript as a file?
Yes, on every plan: Plain Text, Markdown, SRT, WebVTT and JSON, with or without timestamps, for the text you have. While a recording is still being processed, export waits until it finishes. A transcript cut short by the free plan's ceiling exports as the text that exists and is marked as partial.
Does it label who said what?
Yes, when you ask for it. Speaker labelling is a button on a finished transcript, not something that runs by itself: the speaker models are prepared separately from the recognition models, and labelling works on up to 60 minutes of a recording on any plan. On the free plan the labels cover the first five minutes. What it does not change is where the audio comes from: the phone transcribes what its microphone hears in the room, so a remote participant reaches it only through a laptop speaker.
What exactly does the free plan include?
Recording and import are unlimited on it: any number of recordings, any length, and the audio is always saved in full. Each one transcribes its first five minutes, and the speakers are labelled across those five. The higher-accuracy recognition model, Quality, has a free allowance of its own, 15 minutes a month, spent inside those first five minutes, and every one of the 99 recognition languages works whichever model is switched on. Copying and exporting the text you have is free too, in all five formats.
What happens if a call comes in while I am recording?
The call interrupts the recording, and everything recorded before it is saved. Whether recording resumes on its own afterwards is up to iOS, not to the app: if Kaseta shows an error when you come back, it offers Continue and Stop, and you choose. Meeting mode is the reminder to switch on Airplane Mode or a Focus before an important recording, so the call never arrives.
My phone died in the middle of a recording. Is it gone?
The audio that was recorded before the phone died is saved. Open Kaseta, find the recording in the library and check its status: if transcription had not finished, it continues from there. Nothing had to reach a server for that to work – the audio was on the phone the whole time. What no app can recover is sound the phone never recorded, so if you rely on a long recording, start it with a charged battery.
Does the recording stop when my headphones disconnect?
Not by design, but pulling out headphones changes the audio route, and iOS treats that as an interruption. Whether the recording carries on is the system's decision. If Kaseta shows an error, choose Continue or Stop; either way the audio recorded so far is kept. Check the recording status before you trust it on a train.
Can I start recording before the model has finished downloading?
The first launch needs the internet: Kaseta downloads its Fast recognition model and tokenizer before the first recording, and that download is part of setting the app up rather than something you can skip. After it, recording and recognition work offline. The Quality model and the speaker models are downloaded separately, the first time you turn them on.
I left the recorder running by accident. Does it just keep going?
It asks. After a long stretch of silence the app checks whether you are still recording. It does not stop on its own – you decide, so a recorder forgotten in a bag is a question on the screen rather than a decision made for you.
How do I make sure nothing interrupts an important recording?
Turn on airplane mode, or a Focus, before you start. Meeting mode is the reminder to do exactly that: an incoming call is the one thing most likely to interrupt a recording, and preventing it costs one switch. Kaseta needs no network to transcribe once its models are downloaded, so airplane mode takes nothing away from it.
Does it ever make up words it did not hear?
Sometimes, as every speech recogniser does on unclear audio. Check the transcript against the recording – tap a timecode and the audio plays from there – and open a segment to correct it when needed. A stretch of silence can also come back as an empty result rather than as invented words.
There is no text after I stopped. Why?
Check the recording's status in the library. Waiting means it is queued behind another recording, Running means it is still being processed, and an error offers Continue or Stop. On the free plan the text stops at five minutes. A recording made while the models were still downloading waits for them. And a recording of silence can produce an empty transcript – that is the recogniser being honest, not broken.
The transcript came out in the wrong language. What now?
The language is a single setting: Settings, Transcription, Language, where Auto-detect is the default. Pick the language you spoke, open the recording, and tap Retranscribe – the next pass uses that language. There is no language switch on each recording; the setting applies to whatever you transcribe next.
Can I stop a recording without unlocking my phone?
Yes. While a recording runs, a Live Activity sits on the lock screen and in the Dynamic Island with the timer and a stop button, so ending a recording does not mean unlocking the phone and finding the app first.
Can I pause a recording and come back to it?
Yes, pause and resume are both there. You can also drop a flag marker while recording, which is how you find the one moment that mattered afterwards instead of scrubbing through an hour of audio looking for it.
Can I delete the audio and keep the text?
Yes. Storage management lets you remove a recording's audio and keep its transcript, which is what most old recordings eventually deserve: the text is the part you search and reread, and the audio is the part filling up the phone.
Does Kaseta back up or sync my recordings?
No. There is no sync and no backup of its own – the library lives on the phone. Your iPhone's own backup may include it, depending on your iCloud settings. To keep a separate copy of a transcript, export it; an exported transcript is a copy of the text, not of the audio.
How do I find an old recording without listening to all of them?
Search the library: it looks inside the transcripts themselves, not only the titles. Each recording also gets a name of its own and shows the first lines of its text as a preview, so the list is already readable before you search it.
Which iPhones does it run on?
Any iPhone on iOS 17.6 or later. It is an iPhone app only – no iPad or Mac version. How fast a transcript arrives, and whether the Quality model is available, depends on the device.
