Everything runs from your account at surgicalswap.com. This page walks you through logging in, turning on your face, pushing it into a video call, and cloning a voice — start to finish, no guessing.
No links to keep track of — just log in.
About one minute, first time through.
From the dashboard open Start live session. Upload the face you want to become (one clear, well-lit photo), then click Go live. The GPU warms up the first time — give it up to a minute. This must be a computer, not a phone; Chrome or Edge only.
Chrome asks "use your camera?" — click Allow. If you miss it, click the camera icon at the right of the address bar, set it to allow, then refresh. When the preview shows the swap on your face, you're live.
On the live page click Copy OBS link. This copies a private link that carries your swapped video and the synced audio — no screen capturing, no cropping.
In OBS: Sources → + → Browser, paste the link into the URL field, set the size to your camera's (e.g. 1280×720), OK. Your swapped face appears in OBS. Then click Start Virtual Camera in OBS (Tools → Start Virtual Camera). Paste the link once — it stays valid across sessions, so next time you just go live.
In Zoom / Teams / Discord / WhatsApp Web, open video settings and pick OBS Virtual Camera as your camera. Everyone on the call now sees the swapped face.
A separate paid tool on your dashboard — get a clip in any voice.
On your dashboard tap Open voice tool (or go to surgicalswap.com/voice). It uses the same minutes balance as live sessions — 1 minute per clip.
Record a voice sample right there (about 10–30 seconds) or upload a clip of it. This is the voice you want to become — clean audio, one speaker, little background noise gives the best clone.
Record (or upload) up to 60 seconds of you saying the words. Your delivery — the timing and emotion — is kept; only the voice identity is swapped to the sample.
Tick "I own this voice or have permission", then Clone. One minute is taken from your balance; the GPU warms up on the first run. If it can't produce a clip, the minute is refunded automatically.
When it's done the clip plays in the page with a download button. Make as many as you like — each one costs another minute.
Send your converted voice into Zoom, Meet, Teams or Discord.
Your call app can't take the swapped voice straight from the browser. A free "audio cable" sits in between: Surgical plays your converted voice into the cable, and your call app listens to that cable as its microphone — exactly like picking a camera.
Desktop app: on the Voice screen tap Set up voice routing — it downloads and opens the installer for you. Windows (browser): install VB-Cable from vb-audio.com/Cable. Mac: install BlackHole from existential.audio/blackhole. Click through its installer once.
On the live page open Call routing and pick the cable as the output — CABLE Input on Windows, BlackHole on Mac. (The desktop app selects it for you automatically.)
In Zoom / Meet / Teams / Discord audio settings, choose the cable as the microphone — CABLE Output (Windows) or BlackHole (Mac). Everyone on the call now hears your swapped voice.
Use headphones, not speakers. On speakers your mic re-records the swapped voice coming out and it echoes and howls.
Two-minute, one-time setup.
So there are no surprises.
What everything on the page does.
Shape the clone mid-call — no re-preparing.
On the live page, once your voice is prepared, three sliders and a switch appear under the voice picker. They apply instantly, mid-call — drag one and the next second of audio already uses it. Your OBS picture shows the current setting as a small VOICE badge (pitch, strength, level).
The usual fixes, in order.
Allow → refresh. If a phone or iPhone camera keeps getting picked, unplug it or turn off Continuity Camera. Close any other app already using the webcam (Zoom, Photo Booth).Copy OBS link URL. Set the source size to match your camera (e.g. 1280×720).Cmd+Shift+R on Mac, Ctrl+Shift+R on Windows) — the page may be cached. If a slider move prints a message under the knobs instead of changing anything, your voice server runs an older build: Release it and Prepare again for a fresh one.What to click, in order — real screenshots, last step diagrammed.