2025:Symbolic Music Generation Results - Revision history

Junyan: /* Submissions */

2025-09-19T14:23:33Z

‎Submissions

2025-09-19T14:23:14Z

‎Submissions

2025-09-13T05:46:04Z

‎Results

2025-09-13T05:45:20Z

‎Results

2025-09-13T05:44:26Z

‎Results

2025-09-13T05:43:47Z

‎Results

2025-09-13T05:43:23Z

‎Results

2025-09-13T05:22:05Z

‎Results

2025-09-13T05:21:24Z

‎Results

2025-09-12T03:28:34Z

‎Results

@@ Line 13: / Line 13: @@
 |-
 | PixelGen
-| [http://futuremirex.com/portal/wp-content/uploads/2025/symbolic-music-generation/PuxelGen.pdf PDF]
+| [http://futuremirex.com/portal/wp-content/uploads/2025/symbolic-music-generation/PixelGen.pdf PDF]
 | Hierarchical Transformer
 |-

@@ Line 9: / Line 9: @@
 |-
 | RWKV (Zhou-Zheng et al.)
-| [https://www.music-ir.org/mirex/wiki/MIREX_HOME]
+| [http://futuremirex.com/portal/wp-content/uploads/2025/symbolic-music-generation/RWKV.pdf PDF]
 | RWKV
 |-
 | PixelGen
-| [https://www.music-ir.org/mirex/wiki/MIREX_HOME]
+| [http://futuremirex.com/portal/wp-content/uploads/2025/symbolic-music-generation/PuxelGen.pdf PDF]
 | Hierarchical Transformer
 |-

← Older revision		Revision as of 05:46, 13 September 2025
Line 71:		Line 71:
	'''Baseline Models'''		'''Baseline Models'''

−	For MuseCoco, we use the ''xlarge'' model ~~variant~~ with 1.2 billion learnable parameters. For Anticipatory Music Transformer, we use the ''Large'' model ~~variant~~ with 780M learnable parameters.	+	For MuseCoco, we use the ''xlarge'' model with 1.2 billion learnable parameters. For Anticipatory Music Transformer, we use the ''Large'' model with 780M learnable parameters.

← Older revision		Revision as of 05:45, 13 September 2025
Line 71:		Line 71:
	'''Baseline Models'''		'''Baseline Models'''

−	For MuseCoco, we use the 'xlarge' model variant with 1.2 billion learnable parameters. For Anticipatory Music Transformer, we use the 'Large' model variant with 780M learnable parameters.	+	For MuseCoco, we use the ''xlarge'' model variant with 1.2 billion learnable parameters. For Anticipatory Music Transformer, we use the ''Large'' model variant with 780M learnable parameters.

← Older revision		Revision as of 05:44, 13 September 2025
Line 71:		Line 71:
	'''Baseline Models'''		'''Baseline Models'''

−	For MuseCoco, we use the xlarge model variant with 1.2 billion learnable parameters. For Anticipatory Music Transformer, we use the Large model variant with 780M learnable parameters.	+	For MuseCoco, we use the 'xlarge' model variant with 1.2 billion learnable parameters. For Anticipatory Music Transformer, we use the 'Large' model variant with 780M learnable parameters.

@@ Line 67: / Line 67: @@
 Results are reported in the form of mean ± sem<sup>s</sup> (sem refers to standard error of mean), where s is a letter. Different letters within a column indicate significant differences (p-value p < 0.05) based on a Wilcoxon signed rank test with Holm-Bonferroni correction.
 '''Baseline Models'''
 For MuseCoco, we use the *xlarge* model variant with 1.2 billion learnable parameters. For Anticipatory Music Transformer, we use the *Large* model variant with 780M learnable parameters.
 '''Subjective Evaluation Details'''

@@ Line 64: / Line 64: @@
-'''Notes on Evaluation Results'''
+'''Evaluation Results'''
 Results are reported in the form of mean ± sem<sup>s</sup> (sem refers to standard error of mean), where s is a letter. Different letters within a column indicate significant differences (p-value p < 0.05) based on a Wilcoxon signed rank test with Holm-Bonferroni correction.
-'''Notes on Baseline Models'''
+'''Baseline Models'''
 For MuseCoco, we use the *xlarge* model variant with 1.2 billion learnable parameters. For Anticipatory Music Transformer, we use the *Large* model variant with 780M learnable parameters.
-'''Subjective Evaluation Details''': Each test sample was cherry-picked from 8 samples generated from the corresponding prompt. A total of 6 prompts of varied styles (Pop, Classical, and Jazz) were tested, resulting in a 6-page survey. Responses were collected from 20 participants with diverse music backgrounds.
+'''Subjective Evaluation Details'''

@@ Line 63: / Line 63: @@
 |}
-'''Note''': Results are reported in the form of mean ± sem<sup>s</sup> (sem refers to standard error of mean), where s is a letter. Different letters within a column indicate significant differences (p-value p < 0.05) based on a Wilcoxon signed rank test with Holm-Bonferroni correction.
 '''Subjective Evaluation Details''': Each test sample was cherry-picked from 8 samples generated from the corresponding prompt. A total of 6 prompts of varied styles (Pop, Classical, and Jazz) were tested, resulting in a 6-page survey. Responses were collected from 20 participants with diverse music backgrounds.

@@ Line 42: / Line 42: @@
 | 3.58 ± 0.10<sup>a</sup>
 | 3.26 ± 0.10<sup>a</sup>
-| '''3.5 ± 0.10<sup>a</sup>'''
+| '''3.50 ± 0.10<sup>a</sup>'''
 |-
 | PixelGen
@@ Line 65: / Line 65: @@
 '''Note''': Results are reported in the form of mean ± sem<sup>s</sup> (sem refers to standard error of mean), where s is a letter. Different letters within a column indicate significant differences (p-value p < 0.05) based on a Wilcoxon signed rank test with Holm-Bonferroni correction.
-'''Subjective Evaluation Details''': One piece cherry-picked from 8 samples of each test piece, resulting in 6 pages of questions. We collect responses from 22 participants (18 complete submissions and 4 partial submissions). For complete submissions, the average completion time is 16min 59s.
+'''Subjective Evaluation Details''': Each test sample was cherry-picked from 8 samples generated from the corresponding prompt. A total of 6 prompts of varied styles (Pop, Classical, and Jazz) were tested, resulting in a 6-page survey. Responses were collected from 20 participants with diverse music backgrounds.