What is the best voice recorder with transcription?

The one that captures your recordings well, because no transcriber recovers speech the microphone never heard. If you record at a table, in your hand or a metre or two from the speaker, the iPhone you already own is a good enough microphone, and a transcription app on it is the cheapest answer. If you record all day, from the back of a hall, or somewhere a phone cannot sit on the table, a dedicated recorder is the better buy. After that, decide where the text should be made: on your own device, or on someone else's server.

Do AI voice recorders work without internet?

The recording does: a device like Plaud records offline to its own memory. The transcription is another matter. Plaud's happens in its cloud after the device syncs with the phone app, so the text arrives once you have a connection; check how any other model does it before assuming. An on-device app such as Kaseta transcribes on the iPhone itself, so the text appears in airplane mode too.

Can I use my iPhone as a voice recorder with transcription?

Yes. Apple's Voice Memos records well, and recent versions can produce a transcript. Kaseta is built around the transcript instead: record, tap Stop, and the phone turns the recording into text with timecodes, with its speech engine covering 99 languages and nothing uploaded. There is no text while you record – it appears after Stop – and you can export it as Plain Text, Markdown, SRT, WebVTT or JSON on every plan.

Do I need a subscription for transcription?

It depends on where the transcription is computed. A cloud service pays for server time on every minute, so it sells minutes, usually as a plan. On-device transcription costs the developer nothing per minute, so Kaseta does not sell minutes: the free plan gives the text of the first five minutes of each recording and keeps the whole audio, and Pro removes that limit, as a monthly or yearly plan or a one-time purchase.

Voice recorder with transcription – a phone app or a dedicated device?

Search for a voice recorder with transcription and the results mix two different purchases: small gadgets you clip on or set on a table, and apps for the phone already in your pocket. Both record, and both give you text. They differ in where the sound is captured, where it becomes text, and who pays for the computing in between – and those three differences decide which one you should buy.

Kaseta is one of the apps, so read the section on dedicated devices with that in mind: it is the part of this page where we argue against ourselves.

Three things sold under this name

A classic digital voice recorder. A device built to capture audio, with its own microphones and its own battery. Most of them do not turn speech into text by themselves: you move the file to a computer and transcribe it there, with software or a service you choose.
An AI recorder. A small device – Plaud is the name people search – that records offline to its own memory and then sends the audio to the maker’s cloud when it syncs with your phone. The transcript and the summary come back from there, and a subscription pays for the transcription minutes.
An app on the phone. Either a cloud app, which uploads the recording to a server to be transcribed, or an on-device app such as Kaseta, where the speech model runs on the iPhone itself after a one-time download.

When a dedicated device is the better buy

You record all day. A recorder has its own battery and its own storage. A phone recording for hours is still your phone, with its battery, its notifications and its storage shared with your photos.
The speaker is far away. A lecture from the back of a hall or a long boardroom table is a microphone problem before it is a software problem. Dedicated microphones, or a cheap lapel mic, will improve the transcript more than any change of app.
Calls must not interrupt. On an iPhone, an incoming call pauses the recording. Kaseta picks it up again by itself as soon as iOS allows, but the minutes of the call are not on the tape. A recorder on the table does not care who rings.
A phone on the table is not an option. Some rooms do not allow one, and a small dedicated device is a different social object from a phone held out at arm’s length.
You need certificates on paper. Some AI recorder makers hold security audits and certifications for their cloud. An app that sends nothing has no processor to certify, which is a different answer rather than the same one.
You are not on an iPhone. Kaseta is an iPhone app, and that settles it.

When the phone you have is enough

You record where you are. Dictation, an interview across a table, a meeting in one room, a thought on a walk: a phone within a metre or two hears well enough.
The recording should not leave the phone. With on-device transcription the audio is not uploaded – not to be transcribed, not to be stored, not under an account, because there is no account.
There is no network. The speech models live on the phone, so a basement, a plane or airplane mode still produce the transcript after you tap Stop.
You would rather not buy and charge another device. The phone is already in your pocket, and it is the one you will not leave at home.

The questions that actually decide it

Where does the audio go to become text?

With a classic recorder, wherever you take the file afterwards. With an AI recorder, to the maker’s cloud. With a cloud app, to the app’s servers. With an on-device app, nowhere: Kaseta transcribes on the iPhone, and you can check that yourself by putting the phone in airplane mode before you record. How it works is on the offline transcription page.

What does it cost after the first day?

A device is a purchase up front, and an AI recorder adds a plan for the transcription on top of it. An app has no hardware to buy. On Kaseta’s free plan the audio is always saved in full and the text covers the first 5 minutes of each recording. Pro removes that limit and costs $19.99 a year, $39.99 once, or a monthly plan inside the app. There is no minute pack to top up, because the transcription runs on hardware you already own. A side-by-side with a real device and its published plans is on Kaseta vs Plaud.

Does it need a connection?

A recorder records anywhere. Whether the text needs a connection depends on where it is made – see above. Kaseta needs the network once, to download its speech model, and after that recording and transcription both work offline.

Which languages?

Kaseta’s engine transcribes 99 languages. It detects the language from the start of each recording, or you set one in the settings. AI recorders publish their own lists, and the one we compare against on the Plaud page states more languages than ours. A classic recorder transcribes nothing, so its languages are whatever the software you use later can handle.

Battery and time

A recorder spends its battery on recording. A phone that transcribes also does the heavy part itself: long recordings take real time and real battery, and an older iPhone is slower than a new one. Kaseta transcribes while it is open on screen; leave the app and the work pauses, and the app tells you so. That is fine for a meeting’s worth of audio. A back catalogue of a hundred hours belongs on a computer.

Try the phone before you buy the device

The cheapest test costs one recording. Record a real session the way you would use the recorder – same room, same distance, same voices – tap Stop and read the text. If it reads well, you do not need a gadget. If words go missing because the speaker was too far away, that is the microphone, and a different app will not fix it; a lapel mic or a dedicated recorder will. If you also want to know where the audio went, run the same test in airplane mode.

For interviews in particular, interview transcription covers speakers, long recordings and export, and the homepage has the short version of how the app works.