AudioTranscriptor writes out your Italian interviews, lectures and meetings for you. It uploads nothing to the internet, so your recording stays private.
Choose the version for your computer below.
For Mac and Windows · No account · Works offline
Illustrative example: names and figures are made up.
It holds voices, names and things said in confidence. Online transcription services upload the recording to their servers to process it. AudioTranscriptor does the same work on your computer, so nobody else sees the file.
You install AudioTranscriptor like any other program. No account to create.
Put an Italian audio or video file into the window. AudioTranscriptor turns it into text.
Read the text while you listen to the recording, fix what needs fixing and save it.
The first time, AudioTranscriptor downloads the model that recognises speech (about 0.5 GB). After that it works without internet too.
While the recording plays, AudioTranscriptor highlights the sentence you hear. If a word looks wrong, put the cursor on it, press ⌘J (Ctrl+J on Windows) and hear exactly that spot again.
Besides the text, AudioTranscriptor creates subtitles for your videos. If you correct, merge or delete a sentence, the subtitles update on their own and stay in time with the voice.
Save the text as Word (DOCX), plain text (TXT) or subtitles (SRT).
Only two things, and never the recording. The first time, AudioTranscriptor downloads the model that recognises speech. If you turn on English translation, it sends the text, not the audio, to OpenAI with your own personal key. No account and no data on how you use the app.
Try it yourself: turn off Wi-Fi and transcribe a file.
Translation is optional and it is an OpenAI service, not part of AudioTranscriptor. If you want it, you create your own API key on the OpenAI website and enter it in the app's Settings. Before each request AudioTranscriptor shows you how much text goes out and to which model. You pay OpenAI directly for what you use. Transcription stays free.
You turn an hour-long interview into text and copy the quotes, without sending your source's voice to an outside service.
You transcribe interviews and focus groups to analyse. Participants' recordings stay on your computer.
You get the text of a recording to review and save in Word, without it passing through someone else's servers.
From your episodes you get the text and subtitles, already in time after your corrections.
You turn lectures and interviews for your thesis into text, for free, and replay them whenever you need.
The limits, stated up front.
Yes, even if you use it for work. You only pay something if you turn on English translation, and you pay OpenAI directly with your own account.
No. AudioTranscriptor reads and transcribes the file on your computer. Only if you use “Translate to English” is the text (not the audio) sent to OpenAI with your own key.
Create an API key on the OpenAI website (platform.openai.com), copy it and paste it into AudioTranscriptor's Settings. The service belongs to OpenAI: the cost depends on how much text you translate and OpenAI bills you for it. For data sent this way, OpenAI states that it does not use it to train its models and keeps it for at most 30 days for safety checks. If a recording is sensitive, leave translation off.
Only the first time, to download the model that recognises speech: from 0.5 to 3 GB, depending on the one you choose. After that it works offline too. Translation needs internet.
Because AudioTranscriptor has no paid Apple or Microsoft certificate. That does not make it dangerous: the installation guide shows how to open it and how to check the file is the original.
Only if it has an Apple chip, meaning M1 or later. To check: Apple menu, then “About This Mac”, next to “Chip”.
Yes, for your own work. You may not resell AudioTranscriptor or a service based on it. The code is public on GitHub under Apache 2.0 with Commons Clause.
Free, no account. Your computer asks for confirmation on first launch: the guide below shows you how.
Download AudioTranscriptorGet-FileHash FILENAME -Algorithm SHA256shasum -a 256 FILENAME