Building technology to help distinguish authentic human speech from AI-generated audio.
"Somewhere, a parent will answer a phone call and hear what sounds exactly like their child's voice begging for help."
"Somewhere, a family will question whether a final voicemail from a loved one was real."
"The danger of voice deepfakes is not just technological — it is deeply human."
Voice has always been one of humanity's strongest signals of trust. Today, advances in AI make it possible to generate convincing speech at unprecedented scale.
PulmoForge was created to research and develop systems that help restore confidence in digital communication as synthetic voices become increasingly common.
Wilson built PulmoForge's core detection system from the ground up — including the voice liveness architecture, WavLM-based model pipeline, and real-time Zoom audio integration. He leads all technical development and product direction.
PulmoForge started as an independent research effort to understand whether synthetic speech could be reliably detected under real-world conditions — not clean lab audio, but the compressed, degraded audio of actual calls. That question became a product.
Chen leads business development and early customer relationships at PulmoForge. He focuses on bringing the product into the hands of the teams who need it most — working directly with early access partners on onboarding, feedback, and fit.
AI-generated voices will continue to improve. Detection systems must improve faster. PulmoForge is focused on building practical tools, rigorous benchmarks, and research that help people understand what is real and what is synthetic.