Convert audio to text, reliably
Voice notes, meeting recordings, podcasts, conferences: turning audio into text now takes minutes. What matters is knowing how far to trust the result and how to check it without replaying everything.
How automatic transcription works
A speech recognition system listens to the audio and turns it into words, each with its exact start time and a confidence level: how sure it is that it understood correctly. Sbobina Subito uses this to link every word to its minute and to flag the less certain ones.
Which files you can transcribe
- Formats: MP3, WAV, M4A, OGG, FLAC, AAC, OPUS, WMA, WEBM and the audio track of MP4 videos.
- Length: the free first transcript and single transcripts go up to 60 minutes; with a subscription up to 120 or 180 minutes per file.
- Languages: Italian, English, French, Spanish and German.
The file is checked before any payment: if it isn't valid audio or it's too long, it's rejected and you pay nothing.
What accuracy depends on
- Distance from the microphone: the single biggest factor.
- Background noise, echo and music.
- Several people speaking at once.
- Proper names, acronyms and specialist terms the system may not know.
That's why we don't promise a perfect transcript: instead we flag the spots where the system has doubts, so you can check them in a few clicks.
Silences and inaudible passages
A known risk of automatic transcription is "inventing" sentences in silence or noise. Sbobina Subito measures silences directly on the audio: where nobody speaks it writes [silenzio], where speech can't be understood it writes [inudibile] and flags it for checking, instead of guessing.
Fix and download
You fix the text right on the page, with the audio one click away and automatic saving. When it's ready, download it as Word: an appendix lists the uncertain spots you haven't checked yet, with their minute.
Audio and text don't stay on our servers: audio is deleted as soon as you download the Word file and the text after 24 hours.
Step by step
Choose the file
Upload audio up to 60 minutes (longer with a subscription) in any common format.
Set the language
Choose the language spoken in the audio.
Check the text
Review the highlighted spots: one click replays the exact minute.
Download as Word
Download the document; the audio is deleted right after.
Frequently asked questions
How long does an hour of audio take?
Usually a few minutes. We email you when the text is ready, so you can close the page.
What format do I download?
Word (.docx), with title, date and an appendix listing uncertain spots not yet checked.
Is my audio used to train AI models?
No. We use audio only to produce your transcript and choose providers that state they don't train models on customer data.
Can I copy the text from the page?
The text is taken away by downloading the Word file: copy and paste are disabled in the editor.