English: Fast tier
Moonshine Tiny · ~53 MB download
Choose English and Fast for a small initial download. Quality uses the larger Moonshine Base model.
Whisper · 100% in-browser
Free audio to text and video transcription — your audio stays on your device
Turn a recording into text and TXT, SRT, or VTT downloads on your own device. Use Moonshine for English or Whisper for multilingual audio. Approve the first model download, then transcribe locally without an account or per-minute charges.
Audio to text, with subtitle exports
Import a browser-supported audio or video file, choose the spoken language, and transcribe. Review timestamped passages with playback, copy the transcript, or download TXT, SRT, and VTT files. Video decoding depends on the codecs your browser supports.
Start with a short recording
Multilingual: Tiny tier
Whisper Tiny (Quantized) · ~67 MB download
Choose the spoken language in Options and a Whisper tier. Larger models use more memory and take longer to load.
The studio checks device support and shows the download size before approval. CPU fallback can be slower. Start with a short file, check names and punctuation, and keep the tab open while processing. Offline use requires cached model and runtime files.
Generated locally with Kokoro. Download it and choose the file in Transcribe to try the workflow.
“Welcome to the studio. Bring a notebook and a new idea.”
Download sample WAVSide by side
WKO AI vs cloud transcription
Feature WKO AI Cloud services
Price Free, unlimited Pay per minute or capped free tier
Privacy Audio never leaves your device Audio uploaded to cloud for processing
Sign-up None Account required
Languages English via Moonshine · 90+ languages via Whisper, plus Translate-to-English Varies; often English-centric on free tier
Local processing Yes, after a one-time model download No — requires their API
Comparison based on typical hosted transcription services' public pricing and limits; check each vendor for current terms.
Beyond transcripts
Transcribe, synthesize, create — all local
-
English: Moonshine Fast & Quality
English audio defaults to Moonshine Tiny (~53 MB) for the fastest local transcription, with a Moonshine Base Quality tier for difficult recordings — both running in your browser.
-
Multilingual via Whisper
Pick any supported language or Translate to English and Whisper runs locally — Tiny (~67 MB) through Large-v3 Turbo, quantized, no upload.
-
Text-to-speech on device
Reverse the flow with Kokoro 82M TTS — 86 MB, 82M params, 27 neural voices across accents and styles, synthesized in your browser.
Questions
Frequently asked questions
-
Is Whisper transcription really free online?
Yes. The Whisper and Moonshine models run on your own device, so there is no per-minute billing, no free-tier cap, and no sign-up — transcribe as much audio as you like.
-
Do my recordings get uploaded anywhere?
No. Transcription happens entirely in your browser via WebGPU or WebAssembly. Your audio never leaves your device, with no processing upload. Offline use requires all model and runtime assets to be cached.
-
Which languages can I transcribe?
English audio runs on Moonshine for the fastest results, and 90+ other languages run on Whisper — including a Translate to English option for foreign-language recordings.
-
Can I transcribe without an internet connection?
Offline use depends on cached model and runtime files. Complete setup and test transcription before disconnecting.
Transcripts without the upload
Open Transcribe, choose a recording and its language, and approve model setup. Review the transcript and export TXT, SRT, or VTT. Your recording stays on your device.