Drop in an audio or video file and get an editable, word-timestamped transcript — powered by OpenAI Whisper, running entirely in your browser. No upload, no account, nothing your audio ever touches but your own device. Free to run online; a one-time CA$30 Pro add-on unlocks every model, automatic speaker detection, every export format, batch processing, AI summaries, and a fully offline downloadable version.
This is the actual tool — not a screenshot. The frame below loads the Audio Transcriber landing page straight from the server. Scroll it, then jump into the real app below.
Desktop — 1280 px
Mobile — 390 px
Whisper runs as a WebAssembly (or WebGPU-accelerated) model directly in your browser tab. Your audio file is read locally, transcribed locally, and never crosses the network — check the browser's own network tab if you don't believe it.
Drop a filePick a modelTranscribe
The actual app. Drop in any audio or video file, pick Tiny or Base, and watch it transcribe. Nothing is uploaded — it all runs in this browser tab. Best experienced in its own tab.
From opening the page to an editable transcript — no install, no waiting on an upload queue.
Run it online, or double-click audiotranscriber.html — either way, straight into your browser.
Any common audio or video format. It just needs a soundtrack — MP3, WAV, MP4, MOV and more.
Tiny is fastest; Base is a solid balance. Auto-detect the language, or set it directly.
Read along as it streams in, fix any line by clicking it, then copy or download the transcript.
The free version transcribes with the Tiny and Base models — genuinely
useful on its own. Drop audiotranscriber-pro.js next to it and every locked feature
unlocks.
Fast, genuinely useful Whisper models for quick transcripts and voice memos.
Auto-detect, or pick directly — plus one-click "Translate to English."
Word-level timestamps, click-to-edit lines, and playback that follows along as you read.
Copy to clipboard, or download the finished transcript as plain text.
The most accurate Whisper models join Tiny and Base, English-only variants included.
Auto-detect or set an exact headcount, then rename "Speaker 1" to a real name.
.srt and .vtt for subtitles, .json for structured data, on top of .txt.
Queue a whole folder of recordings instead of one file at a time.
A ready-to-run bundle with every model included — zero internet needed, ever again.
Hosted transcription services charge by the minute or by the month, and your audio goes to their servers either way. Here's where a browser-based, local-only tool lands next to them.
| Option | Cost | Model | Your audio |
|---|---|---|---|
| Hosted transcription APIpay-per-use | US$0.10 – 0.40 / minute | Pay per minute, forever | Uploaded to their servers |
| Transcription SaaSsubscription | US$10 – 30 / month | Subscription — stops working if you stop paying | Uploaded, account required |
| Meeting-notes botper-seat subscription | US$8 – 20 / month, per seat | Per-seat subscription | Uploaded to a third party |
| Audio Transcriberruns in your browser — this project | $0 Pro CA$30 once | None — open the page and go | Never leaves your device |
We build fast, self-contained web tools tailored to how you actually work — no templates, no shortcuts.