PR0PH-1T

🙏 @technicallyDE

Went with WebM for recordings. Native support in Chrome and Edge with zero transcoding. Hit record and get a file when you stop. Nothing complicated.

"Just wear a ski mask" ok sure. Until it slips. Or someone recognizes the voice. Or the background. Software handles all of that consistently.

Bandpass filter on audio strips identifying vocal frequencies

Frame rate capped at 15fps so lip reading doesn't work

The anonymization stack in GhostLine:

GhostLine uses WebRTC. After the signaling handshake media flows directly between browsers. Our server couldn't access the stream even if we tried.

Building privacy tools is strange. You're trying to prove a negative. That raw data was never collected. That it can't exist. The only real proof is the architecture.

#HSC2025 with legends.

@PhillipWylie @PhillyInfoSec @HackSpaceCon Thanks!

Attending @BarCodeSecurity’s talk at @HackSpaceCon.

Voice masking in GhostLine: bandpass filter between 400-1500Hz plus a wave shaper and delay node. Strips what makes a voice identifiable but keeps the words totally clear.

Shipped socket reconnection for GhostLine this week. Connection drops mid-interview and picks back up without killing the session. Boring fix but it matters a lot.

Host records on their end. No cloud upload. Raw footage only ever existed in the guest's browser. ghostline.digital

Pixelated video and masked audio go peer to peer over WebRTC. No middleman.

How GhostLine actually works:

If your anonymization runs on a server then the server saw the raw feed. Even for a millisecond that's a copy sitting on someone else's hardware. GhostLine processes locally.