Open-source NVIDIA Broadcast alternative for Linux and macOS video/audio.
Open-source virtual camera and audio tools powered by your NVIDIA GPU.
nvidia-broadcast-linux is an unofficial, open-source alternative to NVIDIA Broadcast built for Linux and macOS users. It brings professional background removal, noise cancellation, and auto-framing to video calls and streams without relying on Windows.
It processes webcam and microphone streams via GStreamer, applying real-time AI background blur, replacement, green screen, auto-framing, and voice enhancement. It also features local Whisper speech-to-text transcription and AI meeting summarization.
It suits Linux and macOS professionals, streamers, and remote workers who want high-end GPU-accelerated video effects and audio noise removal. It does not suit users without compatible hardware or those looking for a zero-config cloud application.
Choose between RVM, IS-Net, and BiRefNet models for background segmentation and edge refinement.
Switch between 9 performance modes ranging from maximum speed to CPU fallback.
Generate GPU-accelerated speech-to-text and local AI meeting summaries with Whisper.
Apply skin smoothing, sharpening, denoise, vignette, and eye contact correction in real time.
Remove microphone and speaker background noise while utilizing customizable voice presets.
Free under its license, with the only cost being the hardware and electricity required to run it yourself.
879 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
GPU-accelerated video inference requires an NVIDIA card, but CPU modes are available as a fallback.
The virtual camera output works with Chrome, Firefox, Zoom, Discord, OBS, and other standard video tools.
Yes, local Whisper transcription and AI meeting summarization run entirely on your local machine.
See Zekai first in Google