-
Creator / Dataset Creator: @safe jolt
-
Extraction method: RMVPE
-
Epochs: 500
-
Dataset length: 04:12 (from available songs on “TYLA” + "Getting Late")
-
Link: https://huggingface.co/spideyyy/ZA_Voices/resolve/main/Tyla.zip
-
Note: From testing the model, I would recommend using RMVPE / RMVPE+ for inference.
Raw acapella vocals work BEST for this model. All audios except for “needy” were inferred with raw vocals; however, processed vocals still work.
My Models
- Brett Cooper (RVC v2) 250 Epochs
- Jacob Zuma (RVC v2) 200 Epochs
