Spaces:
Sleeping
Sleeping
Commit History
-
DP ref encoder uses compressed latent channels from config; README v2 onnx path; pytest smoke tests
Made-with: Cursor8876cf1
-
Simplify uploaded voice style export
Use a single normalized latent stream for both voice style encoders and remove the unused fallback extraction paths so uploaded voices follow one direct exporter flow. Made-with: Cursor35eb04b -
Use root tts config for voice export normalizers
Load the repo root tts.json, split TTL and DP latent normalization during WAV export, and let style extraction fail loudly instead of hiding missing model wiring. Made-with: Cursor6d44c5f -
Extract style_dp from reference WAV via DPNetwork.ref_encoder
Made-with: Cursor7d51a93 -
Clone tab: search fonts/pt_models with filename aliases; vendor models/
Made-with: Cursor863d06f