Audiogen – a new take on generative music AI
Yet another music AI when Suno and Udio already dominate the space.

Dance choreography is genuinely unexplored in AI UGC; quality still early, but the gap exists.
Creators, musicians, content producers, K-pop studios, TikTok creators
Suno · Runway · Kling
So my team built an web app — give it a YouTube link to your music and it generates a 3D dance animation in under 2 minutes. The core is a diffusion-based music-to-motion model (mvnt-m4) trained on proprietary mocap/label data from professional choreographers.
I think dance is the missing piece of AI-generated content — just like how performance made K-pop a global phenomenon, and we believe AI dance will play the same role in the AI UGC era.
This is v0.1 — a fast, experimental playground. Dance quality is still improving (m4.1 in progress), and we're working on faster inference, and finger/facial motion generation. We're also preparing API integrations with platforms like Higgsfield.
I think our tech is quite validated through Epic MegaGrant but still very early in finding user validation. Would love honest feedback on the output quality and what you'd want to see next.
Also in product hunt: https://www.producthunt.com/products/mvntstudio
Demo vid: https://youtu.be/mjq2iAr96iM
Yet another music AI when Suno and Udio already dominate the space.
Viggle AI wrapper specifically for cats, which is genuinely delightful.
Real-time audio diffusion at 25Hz with hot-swappable LoRAs—no engine rebuild needed.
Parallel token decoding beats autoregressive LLMs on throughput, if the math holds up.
npm-style versioning for AI system prompts, but unclear if it solves real pain.
This is a music video, not a software project — wrong venue for Show HN.