Skip to main content
MGToolbox

Voice Activity Detector

AIRuns On Your Device

Find exactly where speech starts and stops in any audio file — Silero VAD runs in your browser, works for every language, no upload.

Real neural voice-activity detection — on your device

This tool runs Silero VAD (~2 MB, downloaded once and cached, works for any language including Mongolian) over your audio file and reports exactly where speech starts and stops. The file never leaves your device.

Typical uses: finding talk time in a podcast or meeting recording, locating segments to cut, or exporting a silence-stripped version.

Select an audio file

Supports MP3, WAV, M4A, FLAC, OGG (Max 200MB)

100% In-BrowserFiles are processed 100% locally
Runs On Your Device
The AI model downloads once and is cached — your files and voice never leave your browser.
Instant Client-Side Speed
Zero network upload delays
Free & Unlimited
No registration or credit card needed

Voice Activity Detector — Where Does the Talking Start?

Upload an audio file and Silero VAD — a genuine neural voice-activity model trained on 6000+ languages — maps every speech segment with timestamps, entirely inside your browser. The file never leaves your device, which matters for meeting recordings, interviews, and voice notes you cannot upload anywhere.

Get the total talk-time ratio, a visual speech timeline, copyable cut points for editing, and a one-click silence-stripped WAV export. The ~2 MB model downloads once and is cached; after that it works offline.

Frequently Asked Questions

Everything you need to know about using Voice Activity Detector safely and efficiently.

No. The VAD model runs inside your browser; the audio is processed on your device and never sent to any server.
Recommended Next Steps

Related Online Utilities