⚡ Fastest is English-only — pick Balanced or Studio Quality for other languages.
Helps recognize names/terms the model wouldn't otherwise know (up to 100 words).
Select transcription quality
Common issues
Everything here runs in your browser, so most issues are fixable on your side — no need to wait on us.
Nothing happens when I click Transcribe
Check that you've picked both a file and a language — the button won't proceed without both.
If that's not it, try a hard refresh: Ctrl+Shift+R (Windows/Linux) or Cmd+Shift+R (Mac). Old cached files can get stuck in your browser after we ship an update.
It's stuck on "Loading AI model…"
The model downloads once per device and needs a stable connection — a slow or interrupted connection can make it look frozen.
Hard refresh and try again. If it still hangs, an ad blocker or privacy extension may be blocking the download — try an incognito/private window to check.
The tab crashed or reloaded during transcription
This is almost always Studio Quality running out of memory — it needs ~6GB+ RAM in-browser. Switch to Balanced or Fastest, or close other tabs to free up memory.
On an older or lower-memory device, cloud transcription (Starter) avoids this entirely — nothing runs on your hardware.
I'm getting an audio decode error
This is a known Safari/iOS bug in the browser's own audio decoder, not usually a problem with your file. Try Chrome or Firefox instead.
It worked before but broke after an update
Your browser is likely holding onto an old cached version of the site. Clear this site's data:
Chrome/Edge: click the icon left of the address bar (🔒 or ⓘ) → Site settings → Clear data. Firefox: click the 🔒 icon → Clear cookies and site data. Safari: Settings → Safari → Advanced → Website Data → find this site → Remove.
Then reload the page.
The transcript looks wrong or repeats words
Try Balanced instead of Fastest — it's meaningfully more accurate and still runs entirely on-device.
Also make sure you've picked the actual language rather than relying on guesswork — telling the model the language directly noticeably improves accuracy over letting it guess.
📂 Recent Drafts
(Saved locally · auto-deleted after 24h)
Everything is free — no account required. Upload audio or video, transcribe, edit, and export.
AudioSRT is a browser-native audio transcription and subtitle generation platform powered by OpenAI Whisper AI models executed via asynchronous Web Workers and WebGPU memory-chunking pipelines. Unlike server-dependent transcription APIs, all media processing occurs entirely on the client device — eliminating upload latency, enforcing zero-knowledge privacy guarantees, and enabling unlimited file duration processing without queue throttling. The platform generates frame-accurate, millisecond-precision SRT subtitle files, plain-text transcripts, VTT caption files, and native Adobe Premiere Pro JSON session files compatible with Premiere's Text-Based Editing workflow. Exported subtitle sequences are fully compatible with Adobe Premiere Pro, DaVinci Resolve, Final Cut Pro X, CapCut, and all SMPTE-compliant NLE systems. The Web Worker architecture completely separates the Whisper inference thread from the browser's primary UI rendering thread, guaranteeing zero UI freezes or tab crashes during multi-hour media file processing.
Transcript will appear here after transcription completes.
Waveform1×
Preview
Scale%
Xpx
Ypx
BG
0:00 / 0:00
Lines
Chars
42
Font
Size
3.5%
Color: #FFFFFF
Style
Opacity
80%
Blur
4px
X
2
Y
2
Width
2px
Color
Position: Bottom
85%
Filler Words
Starter
Detect and remove filler words (um, uh, like, you know...)
from your transcript automatically.
Silence Removal
Starter
Detect and cut silent gaps from your audio and timeline.
Subtitles are automatically re-timed to match.
Min gap:
0.8s
Level:
-40dB
Analyzing audio...
Noise Removal
Starter
Clean up background hiss/hum in the loaded audio — play it back
to hear the difference — then download it. Doesn't change your transcript.
Intensity:
2.5
Processing…
AI Cleanup
Pro
Fix transcription errors and balance subtitle line lengths
using AI.
AI Summary
Starter
Generate a summary and key points from your transcript.
Translation
Starter
Translate your subtitles to any language. This replaces your
current transcript — export first if you want to keep the original.
Viral Clip Finder
Coming soon
Finds the most compelling moments across your whole video and
assembles them into a short story — not just one clipped
moment, a hook-to-payoff narrative pulled from anywhere in
the transcript.
Target length:
60s
Retranscribe
Choose a model to retranscribe your file from the beginning. This will replace the current transcript.
Export transcript
SRT Subtitles
For Premiere Pro, DaVinci Resolve, and any NLE
WebVTT
For web players and HTML5 video tracks
Plain Text
Clean transcript without timestamps
Premiere Pro Starter
Import directly into Premiere Pro Text panel
🎬
Video with Subtitles Starter
Export MP4 with subtitles burned in. H.264 codec, processed locally.