Click record watch the text appear.
Real-time browser transcription using the Web Speech API. No upload, no waiting — start talking and the words land on screen. Best in Chrome; English is the most accurate language.
What you get
What live in-browser transcription gives you.
Zero upload, zero wait
The audio doesn't leave your browser until you choose to save the transcript. The Web Speech API runs the recognition in Chrome itself — start talking, watch the text appear in real time, no server round-trip per word.
For lectures, dictation, notes
Use it as a dictation pad while you write, an in-class note-taker for a lecture you're attending, or a quick capture tool for a thought you don't want to type out. Lower-stakes than a finished file, faster than typing.
Saves to library
If you're signed in, every live session auto-saves to your private library when you stop. If you're signed out, copy the text out of the editor before you close the tab — anonymous live sessions are not retained.
When you need higher accuracy
Live transcription is fast but less accurate than file upload. For a recording you'll quote in print or use for closed captions, record the audio and upload the file — you'll get speaker labels, full punctuation, and 5-10 percentage points of accuracy.
Live transcription vs file upload
When to use which.
✗ Live transcription (this page)
Real-time, browser-based, no upload — but lower accuracy and no diarization. Best for note-taking, dictation, casual capture.
- Real-time, no waiting
- No file upload required
- Chrome only for best results
- No speaker diarization in live mode
- English is most accurate
✓ File upload (everything else on Whipscribe)
Upload an MP3 / MP4 / WAV — get speaker-labelled, word-timed, fully punctuated text with 100+ language support. Slower (2-4 min per hour) but materially more accurate.
- Speaker diarization on every job
- Word-level timestamps
- Full punctuation and paragraphs
- 100+ languages auto-detected
- 5-10 pts better accuracy than live
Sample output
Live transcript. No diarization. As-you-speak.
This is what a typical live session looks like — single speaker, no labels, growing as you talk. Stop the session and you can export to TXT or save to your library.
Export
One transcript. Three clean formats.
Every paid tier exports all three. The free tier exports TXT and SRT.
Plain text
De-ummed paragraphs. Ready to paste.
Show notes
Formatted with chapters and pull-quotes.
Machine-readable
Per-word timing + speaker IDs.
Pricing
Honest pricing, no surprises.
Credits never expire. Upgrade or downgrade any month. Free tier resets daily — no signup, no card.
Free
$0/forever
Try every feature for 30 minutes a day. No card.
- 30 min / day
- Speaker labels included
- TXT + SRT export
- No history retention
Pay-as-you-go
$1/hour
Best for one-off projects. Credits never expire.
- $10 minimum top-up
- Every export format
- 365-day history
- API access
Pro
$8/month
Indie creators. 100 hours / month, all features.
- 100 hours / month
- Clips + every aspect ratio
- Branded captions
- Priority queue
Team
$29/month
Teams. 500 hours / month, shared workspace.
- 500 hours / month
- Shared library
- API + MCP for Claude
- Workspace billing
FAQ
Live transcription questions, answered.
Does it work offline?
No. The Web Speech API still routes audio through Chrome's speech-recognition service for the actual transcription, even though it doesn't transit our servers. If your browser loses internet, transcription stops. For offline transcription, run a local Whisper model — we link a guide in the docs.
Which browsers does it support?
Chrome and Chromium-based browsers (Edge, Brave, Arc) — full support. Safari has partial Web Speech API support; transcription works but accuracy and stability vary. Firefox has limited support and is not recommended. Mobile Chrome on Android works; mobile Safari on iOS is unreliable.
Which languages does it handle?
Same languages as Chrome's speech recognition — about 50 languages. English (US, UK, Australian, Indian) is the most accurate; Spanish, French, German, Mandarin, Japanese, and Hindi are well supported. Less common languages may fall back to lower accuracy or fail to detect.
Does it save to my history?
Yes if you're signed in — every live session auto-saves to your private library when you stop recording, and you can re-open, edit, or export it later. If you're signed out, the session text only exists in your browser tab; copy or download before closing the tab.
Why is file-upload more accurate?
File upload runs Whisper-large-v3 (a much larger model) plus pyannote diarization (speaker labels) plus full-context punctuation. Live transcription uses the browser's lightweight on-the-fly model — fast but materially smaller. Expect 5-10 percentage points better Word Error Rate from a file upload of the same audio.
Can I use it for a meeting?
Technically yes — open this page during a Zoom call and it'll pick up your microphone (and audio leaking from your speakers, depending on the device). For real meeting notes we recommend recording the meeting locally and using our /meeting-notes page — better accuracy, speaker labels, and action-item extraction.
Related
Related tools and pages.
Click record. Watch the words appear.
Try WhipscribeOperated by Neugence Technology Pvt. Ltd. · contact@neugence.ai · Security · Privacy · Terms