Weights is a community platform for training and sharing AI voice and image models, letting anyone create covers and characters in custom styles.
Weights is best known for AI music covers: train a voice model on a few songs, and it can sing other tracks in that voice. The platform extends the same training workflow to image models and text-to-speech, with a social layer where creators publish models, covers, and images for others to remix. Everything runs in the browser, so there’s no local GPU setup โ creators train, generate, and share from one place.
The community is the product’s moat: models trained by creators fill niches mainstream tools ignore, from anime voices to specific singing styles. Browser-side training lowers the technical bar, and the remix culture keeps feeds fresh.
Training runs in the browser from a short audio or image set, and publishing to the community invites remixes. Creators manage their models like a small catalog, and the social feed supplies both audience and quality feedback for iterative improvement.
Governance features let creators manage model visibility and usage, and the credits system keeps heavy training affordable for hobbyists. The platform’s social mechanics also make discovery easy: good models rise through remix activity rather than marketing budgets.
Core features are free with community credits; membership tiers add faster training, more generation time, and premium tools.
Music creators, fan communities, and hobbyists experimenting with AI voices and character art. It fits music creators and fan communities exploring AI voices.
โ Editorial rating ยท ๐ Saves are estimated popularity — Data Sources