Transcribe audio to text with AI in minutes
Upload an audio or video file and get an accurate transcript with timestamps. No subscriptions: pay only for the minutes you use.
Works with
- Voice notes
- Meetings
- Recordings
- Videos
- Podcasts
- Lectures
00:00See the result
From audio to text, line by line
Every sentence is linked to its moment in the audio: click to hear it and fix anything you need.
podcast-interview.mp3
0:30
00:003 steps
How it works
Upload your file
MP3, WAV, M4A, MP4, WhatsApp voice notes… up to 5 hours and 2 GB.
AI transcribes it
Our GPU servers turn speech into text with timestamps.
Review and download
Edit anything you want and export to Word, TXT or subtitles.
00:00Who it's for
For every use case
Students, journalists, podcasters, teams… Pick your use case and see how to get the most out of it.
00:00Why Sonorarium
Everything you need to turn speech into text
Accuracy from a leading model
We run Whisper large-v3 turbo on GPUs, one of the most accurate speech recognition models available. Accuracy varies with the language and the audio quality.
60+ languages
Automatic language detection, or pick it yourself. Great for English, Spanish, Portuguese and many more.
Really fast
An hour of audio is usually transcribed in just a few minutes.
Export to any format
TXT, Word (DOCX), SRT and VTT subtitles, or JSON with timestamps.
Built-in editor
Listen to the audio in sync with the text and fix any word before downloading.
Private by default
Files are encrypted in transit, automatically deleted after 7 days and never used to train models.
00:00Pricing
Pay only for what you use
No subscriptions, no commitments. Credits never expire.
Mini
30 minutes · ≈ 0.5 h of audio
$4/ hour
$2 total
Basic
−25%120 minutes · ≈ 2 h of audio
$3/ hour
$6 total
Standard
−30%300 minutes · ≈ 5 h of audio
$2.80/ hour
$14 total
Pro
−35%900 minutes · ≈ 15 h of audio
$2.60/ hour
$39 total
Business
−42%3,600 minutes · ≈ 60 h of audio
$2.32/ hour
$139 total
How much would you pay?
For reference: a meeting usually lasts 30 to 60 min; a lecture, 1 or 2 h; a podcast episode, about 1 h.
You'd pay
$6
Basic · 120 minutes
≈ $3 per hour of audio
One-time payment · No subscription · Minutes never expire
00:00Formats and languages
Transcribe any file, in any language
00:00FAQ
Frequently asked questions
Can I try it for free?
Yes: you get 15 free minutes when you create an account, no card needed. If you like it, buy a pack of minutes — no subscription and no expiry.
How accurate is the transcription?
We use Whisper large-v3 turbo, which is highly accurate on good-quality audio. Background noise, music or overlapping speakers can reduce accuracy.
Which file formats do you accept?
MP3, WAV, M4A, AAC, OGG, OPUS, FLAC, WMA, AMR, WEBM, MP4, MOV, MKV, AVI and more. Up to 2 GB and 5 hours per file.
How long does it take?
Usually a few minutes per hour of audio. You get the text as soon as it finishes.
Are my files private?
Yes. Audio is stored encrypted, deleted automatically after 7 days and never used to train models. You can delete your account and all your data at any time.
Do purchased minutes expire?
No. Purchased minutes never expire and we only deduct the minutes actually transcribed (rounded up to the minute).
What if a transcription fails?
If it fails, the deducted minutes are refunded automatically.