Algorithm info: VR Arc: 2_HP-UVR + UVR-MDX-NET_Main_427 + kim vocal model fine tuned (old)
Submitted by Bas Curtiz
Version: UVR v5.5.0/Patch_12_22_22_23_44/Win
Advanced VR:
Window Size: 272
Aggression: 10
Crop size: 256
Batch Size: 4
Enable TTA: On
Post-Process: Off
High-End Process: Off
Advanced MDX-Net:
Chunks: Auto
Chunk Margin: 44100
Volume Compensation: 1.035
Denoise Output: On
Spectral Inversion: Off
Based on dca100fb1's suggestion at Audio Separation Discord:
"Anyone tried UVR-MDX-NET Inst 3 + UVR-MDX-NET_Main_427 + 2_HP-UVR?"
Replaced UVR-MDX-NET Inst 3 with Kim's vocal modal cause it scores a bit higher in non-ensembled.
Time Elapsed: 04:08:53
Date added: 2022-12-25 |
