GENDERFLIP.IO
> VIDEO GENDER SWAP_

Upload a monologue. We transcribe it, flip the script to the opposite-sex perspective, swap the speaker's face, and render a fresh 9:16 talking-head video.

fal.ai key ▸ byok.config
Your fal.ai key

Uses your fal.ai account. Each video costs ~$2–7 in fal credits, billed to you. Get a key at fal.ai/dashboard/keys. Format: id:secret

Stored locally in your browser. Sent directly from your browser to fal.ai with each request.
quality.cfg ▸ select tier
upload.exe ▸ drop a monologue
⚠ fal.ai key required

▸ HOW IT WORKS

step_1.txt
1
Upload your video
9:16 monologue works best. Under 60 seconds. Must have an audio track with the speaker's voice.
step_2.txt
2
Add your fal.ai key
Stored only in your browser. Each generation is billed to your fal account (~$2 Fast, ~$4 Standard, ~$7 Cinematic).
step_3.txt
3
Pick a quality tier
Fast for quick previews. Standard is the sweet spot. Cinematic uses Seedance 2 + Sync-Lipsync for the most dynamic motion.
step_4.txt
4
We do the work
Whisper transcribes → Claude rewrites from the flipped POV → face editor swaps gender → ElevenLabs voices it → avatar model lip-syncs → we crop 9:16.

▸ EXAMPLES

coming soon ✦
before_M→F.mp4
M→F
Before
after_M→F.mp4
M→F
After
before_F→M.mp4
F→M
Before
after_F→M.mp4
F→M
After

▸ FAQ

faq.hlp
Why do I have to provide my own fal.ai key?

Each video costs real money to generate (~$2–$7 in fal.ai credits depending on quality tier). By having you use your own key, we don't have to charge you — you pay fal.ai directly. A credit-pack option (no API key needed) is coming later.

Is my fal.ai key safe?

Yes. The key never leaves your browser except to be sent directly to fal.ai with each generation request. It's stored in localStorage on your device. genderflip.io servers see it pass through but don't save it. Still — use a dedicated key for this site and rotate it occasionally.

What's the difference between Fast, Standard, and Cinematic?

Fast uses cheaper, quicker models (~$1–2, 1–2 min). Standard is the default sweet spot — Nano Banana Pro face edit, Eleven v3 voice, OmniHuman 1080p talking head (~$3–4, 2–4 min). Cinematic uses Seedance 2 Pro for cinematic motion then Sync-Lipsync v3 to match the mouth to the new voice (~$6–8, 4–8 min) — most watchable but also the slowest.

Why might the face-swap step refuse my video?

Nano Banana Pro (Google's Gemini model) sometimes blocks edits of faces it recognizes as real people. When that happens the app automatically falls back to Seedream 4 which has no such restriction. You'll see a log line and the cost breakdown updates. Fallback adds ~$0.03.

What length of video works?

Monologues under 60 seconds work best. The avatar model at 1080p caps at ~30 seconds of output audio per request. Longer inputs are technically fine but the cost scales linearly ($0.16/sec for the avatar step alone).

Can I use this on my own video?

Yes — that's the main use case. The more clearly framed the speaker (centered, good lighting, one face), the better the swap. Low-res or heavily stylized inputs produce lower-quality swaps.

Can I use this on a celebrity or public figure?

You can — some face-edit models refuse, others don't. Use the tool responsibly: for satire, parody, and perspective-flipping commentary. Don't misrepresent real people's actual statements as authentic, and don't use the output to harass anyone. You're responsible for what you make.

Where are my uploaded videos stored?

On the server's local disk for ~1 hour during processing, then deleted. Generated outputs are served via a signed URL and expire on the same cleanup cycle.