Seed-VC checkpoints (mirror)

An unmodified, file-for-file mirror of Plachta/Seed-VC, taken 2026-08-26. All credit to Plachta (github.com/Plachtaa/seed-vc); weights are GPL-3.0 as licensed upstream (full text in LICENSE).

Contents: the V1 DiT checkpoints (DiT_seed_v2_uvit_whisper_small_wavenet for speech, the whisper_base_f0_44k variants for singing, DiT_uvit_tat_xlsr tiny), the V2 stack (v2/ar_base.pth, v2/cfm_small.pth), the hift vocoder and configs.

Mirrored for use in a local macOS music app whose voice-conversion entries fetch these at first run — this copy is archival insurance against the original (a personal account) going away; the app downloads from the original while it exists. The original card carried no prose, only the license front-matter reproduced above.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support