TXTTranscriptionTranscribe audio to text
One model: Whisper on our own servers. It does not tell speakers apart: the text comes in short paragraphs, each with its timestamp.
Frequently asked questions
Transcribe audio to text
How do I transcribe a WhatsApp voice note?
On your phone, share the voice note and save it as a file: on an iPhone with «Save to Files», on Android in any folder you like. Then upload it here. WhatsApp voice notes are .opus files and we read them as they are, with no conversion needed first.
How much does it cost to transcribe audio?
Up to 30 minutes of audio it is free and needs no account. Longer audio is paid once, online and still without an account: 1 € up to 60 minutes, 2 € up to 120, VAT included, with the invoice by email. You see the price before paying. If the transcription fails we refund you without you having to ask.
Which files can I upload?
MP3, M4A, WAV, OGG, FLAC and AAC audio, WhatsApp voice notes and MP4, MOV, WebM and MKV video. From a video we keep only the audio, and your browser does that before uploading, so the file is much smaller. You can also record from the microphone right on the page. The maximum is 120 minutes.
Which languages does the transcription work in?
The model knows 99 languages and finds the language of the audio by itself. You can also pick it yourself: useful when the audio starts with music or noise, because detection only listens to the first seconds. English, Italian, Spanish, French, German and Portuguese come out the most accurate; less common languages have more mistakes.
Does it tell who is speaking?
No. The model we use does not tell voices apart, and we do not pretend it does. The text comes in short paragraphs, each with the timestamp it starts at, so you can find a point even without names. For a meeting there is Meeting Minutes AI, which turns the transcript into a summary, decisions and action items.
How long does it take to get the text?
It depends on the length of the audio and on how many transcriptions are ahead of yours: the server does one at a time. While you wait, the page shows how much time has passed and, when there are enough recent transcriptions to work it out, how long it usually takes. You can switch tabs: the work keeps going.
Which formats can I download the text in?
Copy it with one click, or download it as TXT, Word (DOCX) or PDF, with or without timestamps beside the paragraphs. There are also SRT and VTT subtitles, which open in YouTube, Premiere, CapCut and VLC. On the page you can search for a word and, if you chose the audio here, click a timestamp to listen to that point again.
Where do the audio and the text go?
The audio goes to our European servers, is transcribed there and is deleted as soon as the work is done. It does not pass through other companies' transcription services. The text stays for 24 hours, so you can reopen it from this browser or from the link in the payment email, then we delete it. We do not use it to train any model.