Last month I got an 18-minute WhatsApp voice note from a client explaining a whole project — the kind you can't listen to right then because you're in the middle of something else. I just wanted to read it. I went looking for an online transcriber and hit the same wall as always: either the site wanted me to upload the audio (a private client briefing!) to a stranger's server, or it was paid, or it capped me at 2 minutes. I closed all the tabs, and that weekend I built AudioWriter.
The problem: transcribing audio is tedious — or invasive
When you need to turn audio into text — a voice note, a recorded meeting, an interview, a lecture — the usual options are all bad in one way or another:
- Listen and type by hand. It works, but 18 minutes of audio becomes an hour of pausing and rewinding. Nobody has that time.
- Upload the audio to a site. Most online transcribers send your file to a server to process it there. Fine for a meme — but what about a confidential meeting, a personal conversation, a client's audio? You're handing it to a third party you don't know.
- Pay or hit a limit. Many charge per minute or cut you off after a snippet — right when the audio is long and you need it most.
- Convert the format first. WhatsApp audio comes as .ogg, and half the tools won't take it — so you have to convert first, one more chore.
The common thread: you just want the text, and the road costs you either your time or your privacy.
The solution: AudioWriter, and the audio never leaves your device
AudioWriter does this: you drop the audio and it transcribes it inside your browser. That's the whole point — the AI (Whisper, OpenAI's speech model) runs right on your machine, so the file isn't sent to any server, doesn't go through an API, isn't stored anywhere but your own computer. It's free, no signup, and takes WhatsApp's .ogg directly (plus MP3, WAV, M4A).
The first time, it downloads the AI model once (cached after that); from then on the transcription runs even offline. For "simple transcription" — the tool's focus — that's exactly what matters: drag, wait, read.
What it gives you
- Real privacy (not a slogan). Since the processing is in the browser, you can transcribe a meeting, an interview or a sensitive conversation without handing the audio to anyone.
- Takes WhatsApp audio. Drag the voice note's .ogg/.opus and you're done, no converting first.
- Editable text + .txt + .srt subtitles. You review the text, copy it, download a .txt — or generate an .srt subtitle file with timestamps, ready to drop into a video. That's the trick for anyone making content.
- Fast or accurate, your call. A light, fast model by default; a more accurate (heavier) one in a click, when quality matters more than speed.
- Live dictation as a bonus. There's also a real-time dictation mode for quick notes. Honestly: that mode uses the browser's speech engine (which in Chrome may go through the cloud) — the 100%-local promise is for the file mode, which is the heart of the tool.
FAQ
Do I need to install anything or create an account? No. It's free, no signup, runs in the browser — open it, drop the audio, get the text.
Is my audio sent to any server? In "Transcribe file" mode, no: the AI runs in your browser and the file isn't uploaded. "Live dictation" uses the browser's speech engine (in Chrome it may go through the cloud) — so the 100%-local promise is for file mode.
Does it work with WhatsApp audio? Yes, it takes the voice note's .ogg/.opus, plus MP3, WAV and M4A — just drag it in.
Does it need internet? Only the first time, to download the AI (then it's cached). After that, file transcription runs offline.