Skip to main content
EURASIP Journal on Audio, Speech, and Music Processing

Table 3 MUSIB evaluation on JSB Chorales Dataset

From: MUSIB: musical score inpainting benchmark

JSB Chorales Dataset (\(\approx\)2.4K samples)

Model

\(NLL \downarrow\)

\(pos_{F1} \uparrow\)

\(pAcc \uparrow\)

\(rAcc \uparrow\)

\(S_{div} \downarrow\)

\(H_{div} \downarrow\)

\(GS_{div} \downarrow\)

Anticipation-RNN

0.459

0.832

0.243

0.682

0.240

0.525

0.232

InpaintNet

0.327

0.852

0.505

0.788

0.059

0.411

0.153

SketchNet

0.605

0.833

0.272

0.708

0.079

0.529

0.228

VLI

0.935

0.643

0.163

0.639

0.197

0.585

0.225

  1. Values in boldface highlight the best performance among the four models on each metric
Back to article page