Audio to text free
Turn WhatsApp voice notes, interviews, lectures, meetings or videos into text — no minute limit, no sign-up and no upload. Click any sentence and the audio jumps to it.
Your audio never leaves your device: transcription happens in your browser.
Drop your audio or video here
or click to choose · MP3, WAV, M4A, OGG, OPUS, MP4, MOV, WebM · nothing leaves your device
More video and audio tools
How to transcribe audio to text
Drop the file, check the audio language and tap Transcribe. The text appears every half minute of audio, in paragraphs, with a timestamp for each part. Click any sentence and the player jumps to that moment; press play and the part being spoken is highlighted.
Use search to find a word: each result takes you to the moment it was said. When it’s done, download it as TXT, DOCX (Word and Google Docs) or PDF — with or without timestamps — or as SRT and VTT subtitles for YouTube, VLC and video editors.
Transcription is done by Whisper, OpenAI’s speech recognition model, running on your own computer’s graphics chip. The first time, your browser downloads the model (564 MB); after that it’s stored. On an ordinary laptop, an hour of audio takes about 45 minutes, and you can switch tabs meanwhile.
No account and no limits: the text is kept only in this browser, and if the tab closes halfway through you can pick up where it stopped. To come back to a transcript another day, drop the audio together with the .srt or .vtt you downloaded. Need to pull the audio out of a video first? Use Video to MP3.
Example: the interview that becomes an article
A 4 min 37 s interview was transcribed in 3 min 17 s on a laptop with integrated graphics. With the text ready, searching for a name jumps straight to where it was said, clicking a sentence plays that moment so you can check the quote, and the DOCX download with timestamps goes into your text editor. To caption the interview video, the same text comes out as SRT.
Frequently asked questions
Is it really free? Is there a minute limit?
It’s free, with no sign-up and no limit on minutes or files. The only limit is your device: on a computer, hours-long recordings work; on a phone it’s slow and may not run.
Is my audio uploaded to a server?
No. The audio is read and transcribed inside your browser, on your device’s graphics chip. Only the transcription model is downloaded, once, from Hugging Face; your file goes nowhere.
How do I transcribe a WhatsApp voice note?
Save the voice note on your device (on Android, through Share; on iPhone, with Save to Files) and drop the .opus or .ogg file here. Pick the language and tap Transcribe.
How long does it take?
It depends on the computer. On a laptop with integrated graphics, a 4 min 37 s interview took 3 min 17 s, about 70% of its length. The screen shows the time left, measured on your own device.
Why does it download 564 MB the first time?
That’s the Whisper model, which does the speech recognition on your device instead of sending the audio to a server. It’s downloaded once and stored in the browser: from the second time on, it opens in seconds.
Does it work with video?
Yes. Drop an MP4, MOV or WebM: the tool takes the audio and transcribes it. The video shows in the player, and clicking the text moves the video to that moment.
Can it make subtitles?
Yes: download SRT or VTT. Both are subtitle formats with a timestamp for each line, accepted by YouTube, VLC and video editors.
Does it work on a phone?
Transcription uses the graphics chip through the browser (WebGPU). On phones it’s slow and may not run; on a computer with an up-to-date Chrome or Edge it works well.
Which languages can it transcribe?
Dozens, including English, Portuguese, Spanish and Indonesian. Choose the audio’s language before transcribing: the result is more accurate than letting the model guess.
Guides to get your video and audio right
WhatsApp voice notes, MP4 to MP3, smaller videos, formats, codecs and metadata.
Transcription with Whisper large-v3 turbo (OpenAI, MIT license), in the onnx-community ONNX build, running in your browser through transformers.js (Hugging Face, Apache-2.0) and your device’s graphics chip (WebGPU). The model is downloaded from Hugging Face the first time; your audio never leaves your device.