How to transcribe audio to text: 4 methods compared

There are four ways to transcribe audio to text: type it out by hand, use your phone’s built-in transcription, upload the file to an online AI tool, or import it into a meeting notes app like qatip. A two-minute voice memo needs nothing more than your phone. An hour-long meeting with four speakers is where AI saves your evening.

What should you decide before you start?

The right method depends on what you’ll do with the text. A few questions up front keep you from wasting an hour on the wrong tool:

• Do you need a word-for-word transcript, or just a clear record of what was said?
• How many people are talking, and does it matter who said what?
• What language is the recording in? Do people switch languages mid-sentence?
• How long is it?
• Does it contain anything sensitive, like client details, health information or salaries?

People skip that last one most often. When you upload a file to a website, the audio sits on another company’s servers, and how long they keep it is up to their rules, not yours.

1. Transcribing audio by hand

This is the oldest method, and still the most reliable, because you hear and type every word yourself. The cost is time. Anyone who has typed out an hour of audio knows it takes several times longer than the recording itself.

If you’re doing it by hand, work in this order:

1. Open the file in a player that can slow playback down, and set it to 0.75x.
2. Work in short chunks: listen for a few seconds, pause, type.
3. Start a new line whenever the speaker changes, and put their name or initials first.
4. Don’t guess at words you can’t make out. Mark the timestamp instead, like [12:40 unclear].
5. When you’re done, listen once more at normal speed and read along.

For interview quotes, legal work or research, where every word has to be right, it’s worth doing this final read-through even when a tool wrote the first draft.

2. Your phone’s built-in options: iPhone and Android

The recording app already on your phone may do part of the job. Check which languages it supports before you rely on it.

iPhone: Voice Memos and Notes

Voice Memos can transcribe recordings. While you record, swipe up on the waveform and tap the transcription button to watch the text appear live. Afterwards, open the recording, tap More, and pick View Transcript or Copy Transcript. If you record audio inside a note in the Notes app, you get a transcript there as well.

Apple’s support page spells out the limits. You need an iPhone 12 or later, and transcription works in English, Spanish, Portuguese, Italian, French, German, Japanese, Korean, and Simplified and Traditional Chinese. It isn’t offered in every country or region. On Apple Intelligence devices, Writing Tools can summarize the transcript too.

Android: Pixel’s Recorder app

The Recorder app on Google Pixel phones transcribes while you record. The live transcription covers a short list of languages. For a saved recording, the Re-transcribe option supports a much longer list of languages and regional variants. Google notes that Re-transcribe may process audio files on its servers. Speaker labels currently work only in US English, and summaries need a Pixel 8 or later.

On a Samsung or another brand, check that phone’s own recorder app, because features differ from one maker to the next. Google’s Live Transcribe app turns speech into text in real time, but it’s built for showing what’s being said right now, not for transcribing a file you already have. It keeps your transcription history for up to three days.

3. Online AI transcription tools

There are plenty of sites where you upload a file and get text back a few minutes later. Most of them do the same thing under the hood: they pass the audio to a speech recognition model and return the result. The differences are in the details, so check these before you pick one:

• How well does it handle your language? Test it with a short clip from one of your own recordings, ideally one where people mix in jargon or a second language.
• Does it separate speakers, or hand you one long block of text?
• Do you only get a transcript, or also a summary and a list of next steps?
• Where is your file stored, for how long, and is it used to train AI models? If the privacy page doesn’t say, don’t upload.
• How many minutes are free, and what does it cost after that?

These tools work well for a one-off file. If you transcribe meetings every week, though, the results end up scattered across downloads, and you can’t search across them.

4. Importing the file into qatip (phone and Mac)

If you already have a recording, say a meeting sitting in Voice Memos or an MP3 a colleague sent you, you can import it into qatip and get the same transcript and report you’d get from recording with qatip.

1. On your phone, open qatip and use the import option to pick the audio file from Files. On a Mac, use the Import button.
2. The file uploads, and a few minutes later the transcript is ready, split line by line by speaker.
3. The report comes with it: a summary, decisions, to-dos with an owner and a date, and open questions.
4. The recording sits next to your other meetings, so you can search for something that was said or ask a plain question across all of them.

You don’t need to pick a language. qatip detects close to 100 languages on its own, and handles meetings where people switch between two languages. Importing works on the phone and on Mac. The Windows app doesn’t have it yet.

The free plan includes 300 minutes a month, with the first 30 minutes of each recording written up. For longer recordings or more minutes, see the pricing page.

Which method should you use?

Method	Languages	Speaker separation	Summary and to-dos	Best for
By hand	Any you understand	You add it	You write it	Short recordings that need every word right
iPhone Voice Memos and Notes	10 listed languages	Not mentioned on Apple’s page	Summary with Apple Intelligence, on supported devices	Quick memos in a supported language
Pixel Recorder	A long list for saved files, a short one live	US English only	Summaries on Pixel 8 and later, limited languages	Everyday recordings on a Pixel
Online AI tools	Varies by tool	Varies by tool	Varies by tool	One-off files
Importing into qatip	Close to 100, detected automatically	Yes, line by line	Yes: summary, decisions, to-dos	Regular meeting recordings with several speakers

How do you get a cleaner transcript?

Whatever method you choose, most of the quality comes from the recording itself. Put the phone in the middle of the table, close to everyone. Keep it away from anything that hums all the time, like an air conditioner, a coffee machine or an open window onto a street. If everyone says their name once at the start, telling voices apart later is much easier. And when two people talk over each other, no tool will get that moment fully right.

If you record an in-person meeting with qatip from the start, you skip the import step altogether, and it keeps recording with the screen locked. If transcribing recordings is a regular part of your week, the audio to text page shows how it works.