On clean English or Spanish audio we sit around 98% word-for-word accuracy. With background noise, several people talking over each other or strong accents, the usual range is 92% to 96%. You can always review and fix things in the editor.

Which languages do you support?

More than 90 languages and variants, including English, Spanish (European and Latin American), Catalan, Galician, Basque, French, German, Italian and Portuguese. Language detection is automatic, and we handle recordings that switch language partway through.

What happens to my data?

Your files are processed and stored in European Union data centres, encrypted in transit and at rest. We never use your content to train models. You can permanently delete any recording in one click and, on paid plans, set automatic deletion after 7, 30 or 90 days.

Can I cancel whenever I want?

Yes. Billing runs on Stripe, and you can cancel or change plan from your billing page at any time. If you cancel you keep access until the end of the period you already paid for, and your transcripts stay available to download.

Not yet. We're building a REST API with webhooks so you can transcribe from your own code, and it will be included in the plan at no extra cost. Write to us and we'll tell you when it opens — and what you need it for, which is how we decide what to build first.

Which file formats do you take?

MP3, WAV, M4A, AAC, FLAC, OGG, MP4, MOV, WEBM and MKV, plus public links from YouTube, Drive or Dropbox. The per-file limit is 1 GB on Starter and 10 GB on every other plan.

audio2text.org

Transcribe audio and video
into text in seconds

Fast, accurate AI transcription and translation. Drop your file in and read the text before your coffee gets cold.

Drag your file here

MP3, WAV, M4A, MP4, MOV… up to 10 GB

MP3
WAV
M4A
AAC
FLAC
OGG
OPUS
OGA
WMA
AMR
AIFF
AIF
CAF
MPGA
M4B
DSS
DS2
MP4
MOV
WEBM
MKV
AVI
M4V
WMV
FLV
MPG
MPEG
3GP
3G2
MTS
M2TS
TS
OGV

Companies and freelancers who trust us with their recordings

KomboRadio FábricaVell GroupHelpstackNordwindStudio VelaAtlas LegalMediterraKomboRadio FábricaVell GroupHelpstackNordwindStudio VelaAtlas LegalMediterra

Features

Everything that comes after you hit record

It isn't just audio turned into text: it's a recording you can actually work with.

Our own engine

98% average accuracy, background noise and accents included

We pair an acoustic model trained on real meeting audio with a contextual pass that knows industry jargon, proper nouns and acronyms, so «Kubernetes» stops coming back as «Cooper Netties».

98%: accuracy
<2 min: per hour of audio
90+: languages

00:41 · confidence 0.98

Speaker diarization

Who spoke and when, worked out automatically even when voices overlap. Name each person once and it applies to the whole transcript.

More than 90 languages

English, Spanish, Catalan, German, French… detected automatically, and we handle recordings that switch language halfway through.

Summaries and minutes

Key points, decisions and action items pulled from the meeting, each one linked to the exact minute it was said.

Export to whatever you use

DOCX, PDF, SRT, CSV and JSON, with timestamps. Or copy straight to the clipboard.

Search by meaning

Search your whole archive for what was meant, not just what was literally said, and jump to the moment it happened.

API and webhooks · in development

A documented REST API and webhooks so you know the moment a job is done, and can transcribe from your own code. Not open yet: we're finishing it and will email you as soon as it's usable.

Private by default

Encrypted in transit and at rest, hosted in the EU, automatic deletion on your schedule and zero training on your data.

Genuinely fast

An hour of audio is done in under two minutes. No waiting in line on paid plans.

How it works

Three steps and you're done

Nothing to install, no models to configure, no documentation to read before you start.

01
Upload or connect
Drag a file in, paste a link to a recording, or record it right in the browser. From there it goes straight into the queue.
02
Let it work
We detect language and speakers, transcribe with timestamps and write the summary. An hour of audio, under two minutes.
03
Edit and export
Fix anything in the editor synced to the audio, share it with your team and export in the format you need.

Real example

This is what a 54-minute interview looks like

An untouched excerpt from a real product research interview.

00:11 / 02:02

Customers

Teams that stopped transcribing by hand

We went from spending two days transcribing research interviews to having them ready before lunch. The diarization gets who's speaking right even when we talk over each other.

Nuria Alcaraz

Head of Research, Kombo

We subtitle eight podcast episodes a week. The timestamped editor saved us from hiring someone just for that.

Dani Sáez

Producer, Radio Fábrica

The automatic minutes from our committee meetings are surprisingly good. We review them now instead of writing them.

Elena Bosch

Chief of Staff, Vell Group

We wired the API into our support pipeline in an afternoon. The webhooks behave exactly like the docs say they do.

Marc Tudela

CTO, Helpstack

2.4M: hours transcribed
98%: average accuracy
90+: languages
12,000: teams

Pricing

Three plans. Only the minutes change

They all do the same. Start free with 60 minutes a month and move up when you record more. No commitment: cancel whenever you like.

You're charged in the currency you choose.

Free

Try it with your own recordings, no card needed.

$0/mo

60 minutes a month

one hour of audio

One free account per person and device.

Create a free account

Included in all three plans

AI transcription in 90+ languages
Speaker separation and summaries
Translation into 86 languages
Export to DOCX, PDF, SRT, CSV and JSON
Files up to 10 GB
Search inside the text

Subscriptions renew every month until you cancel, and you can cancel any time from your dashboard. Prices exclude VAT. Need more than 8,000 minutes a month or to pay by bank transfer? Talk to us.

FAQ

What people ask us most

Not here? Write to us and you'll hear back the same working day.

Try your first recording without paying

Upload it and read the first lines without a card. A free account gives you 60 full minutes a month.

Create account See plans

Transcribe audio and videointo text in seconds