Phonic-1 comes from a 10-person MIT/Stanford/CMU team rebuilding the voice stack
graceisford · x · 2026-10-30
Phonic introduces its 10-person team of researchers and engineers from MIT, Stanford, and CMU, including International Olympiad medalists and former founders. Their thesis: building at the frontier means working deeply with customers, since voice's hardest problems only appear in real calls—leading them to rethink every layer of the stack (data, pre-training, RL, kernels, infrastructure) into Phonic-1, a model that doesn't trade intelligence for naturalness.
Related event: Phonic Launches Phonic-1 Speech-to-Speech Model with $27.5M Series A(7 posts)→
More from Multimodal
- Medusa uses nine synced iPhone 18 Pros and four Mac Studios to build reframable 3D scenes — SumitGup · 2026-10-30
- Holo4D accepted to NeurIPS: reshoot any video from a camera that was never there — HirokatuKataoka · 2026-10-30
- WanGP one-click launcher hits Pinokio: AI video generation on 6GB+ VRAM GPUs — cocktailpeanut · 2026-10-30
- Wan2GP's new Deepy Prime turns one idea into games, stories and 3D scenes with generated assets — cocktailpeanut · 2026-10-30
- Simulon turns any photo into a photoreal 3D asset — user brings an extinct quagga back to life — TinfoilTricorn · 2026-10-30
- Cinematic title sequence built with GPT Image, Seedream, LTX and MiniMax-H3 — magratheya_64 · 2026-10-30