Skip to content
AI Lab

Where We Explore What AI Makes Possible

The Lab is our R&D playground: generative video, voice cloning, gaussian splatting, lip-sync, speech intelligence. Some experiments stay experiments. Some graduate into production systems — like the virtual persona now answering 50+ questions a day.

See what graduated
Explore
Working Prototype
01
Video · Lip-Sync

AI Video Dubbing & Lip-Dub

Our own end-to-end pipeline: breaks video into chunks, transcribes with word-level alignment, translates to any language, clones the original speaker's voice, and lip-dubs the result back onto the footage. GPU-heavy stages run as serverless workers.

WhisperXNLLB-200CosyVoice3LatentSync
Working Prototype
02
Video · 3D

DeepSpeech AI Avatar

The next step past dubbing: a photorealistic gaussian-splat talking head driven by a cloned voice — an AI presenter generated from a few minutes of real footage, speaking any language.

GaussianTalkerDeepSpeechVoice cloning
Working Prototype
03
3D Capture

Gaussian Splat 3D Tours

A 5-minute phone walkthrough becomes a navigable, photorealistic 3D tour — built for real estate, hotel rooms, and amenities. Our own desktop app handles splat generation and editing; one command runs the whole video-to-tour pipeline.

3D Gaussian SplattingCOLMAPRustPhone capture
Active Experiment
04
Real-Time Graphics

GPU Compute Simulation

Compute-shader experiments in Rust: hundreds of thousands of particles and voxels simulated and rendered in real time.

RustwgpuCompute shaders
In Production
05
Speech

Speech Intelligence

Speech-to-text with word-level timestamps, speaker diarization, and translation — the foundation layer under our dubbing pipeline and voice products.

WhisperXDiarizationAlignment

Have a Wild Idea?

The Lab exists because clients kept asking "is this even possible?" Usually the answer is yes. Bring yours.

Book a Free Call