2015:Music/Speech Classification and Detection Results

Introduction

These are the results for the 2015 running of the Music/Speech Classification and Detection task. For background information about this task set please refer to the 2015:Music/speech classification and detection page.

General Legend

Sub code	Abstract	Contributors
GWDS1	PDF	Vikaskumar Ghodasara, Daimi Syed Naser, Shefali Waldekar, Goutam Saha
JS2	PDF	Jan Schlüter
JS3	PDF	Jan Schlüter
MM1	PDF	Matija Marolt
MM2	PDF	Matija Marolt
MM3	PDF	Matija Marolt
RHM1	PDF	Jimena Royo-Letelier, Romain Hennequin, Manuel Moussallam
RS2	PDF	Reinhard Sonnleitner
TL1	PDF	Thomas Lidy
UK1	PDF	Aiko Uemura, Jiro Katto
ZC1	PDF	Chaogang Zhang, Chuan-Yi Chen
ZC2	PDF	Chaogang Zhang, Chuan-Yi Chen
ZC3	PDF	Chaogang Zhang, Chuan-Yi Chen
ZC4	PDF	Chaogang Zhang, Chuan-Yi Chen
ZC5	PDF	Chaogang Zhang, Chuan-Yi Chen
ZY1	PDF	Wei-Qiang Zhang, Xu-Kui Yang
TVDP1	PDF	Nikolaos Tsipas, Lazaros Vrysis, Charalampos Dimoulas, George Papanikolaou
TVDP2	PDF	Nikolaos Tsipas, Lazaros Vrysis, Charalampos Dimoulas, George Papanikolaou
TVDP3	PDF	Nikolaos Tsipas, Lazaros Vrysis, Charalampos Dimoulas, George Papanikolaou
NT1	PDF	Nikolaos Tsipas, Lazaros Vrysis, Charalampos Dimoulas, George Papanikolaou
NT2	PDF	Nikolaos Tsipas, Lazaros Vrysis, Charalampos Dimoulas, George Papanikolaou
YZ1	PDF	Wei-Qiang Zhang, Xu-Kui Yang

Task 1: Music/Speech Classification

Complete dataset

Sub code	Accuracy	std
GWDS1	0.9866	0.0065
JS3	0.9946	0.0027
MM1	0.9869	0.0037
MM2	0.9754	0.007
RS2	0.9962	0.0029
TL1(15)	0.9927	0.0018
TL1(80)	0.9973	0.0018
ZC4	0.9429	0.0076
ZC5	0.9429	0.0076
NT1	0.9835	0.0109
NT2	0.9935	0.0044
RHM1	0.9954	0.004
YZ1	0.9835	0.0044

Individual Results Files for Task 1

GWDS1= Vikaskumar Ghodasara, Daimi Syed Naser, Shefali Waldekar, Goutam Saha
JS3= Jan Schlüter
MM1= Matija Marolt
MM2= Matija Marolt
RS2= Reinhard Sonnleitner
TL1(15)= Thomas Lidy
TL1(80)= Thomas Lidy
ZC4= Chaogang Zhang, Chuan-Yi Chen
ZC5= Chaogang Zhang, Chuan-Yi Chen
NT1= Nikolaos Tsipas, Lazaros Vrysis, Charalampos Dimoulas, George Papanikolaou
NT2= Nikolaos Tsipas, Lazaros Vrysis, Charalampos Dimoulas, George Papanikolaou
RHM1= Jimena Royo-Letelier, Romain Hennequin, Manuel Moussallam
YZ1= Wei-Qiang Zhang, Xu-Kui Yang

Task 2: Music/Speech Detection

Sub code	fb_F	eb_F_500ms	eb_Foff_500ms	eb_F_1s	eb_Foff_1s
JS2	0.8233	0.1754	0.1212	0.2387	0.1639
MM3	0.8941	0.4037	0.2911	0.4419	0.3085
TVDP1	0.8314	0.1490	0.1146	0.2233	0.1718
TVDP2	0.8346	0.1250	0.1231	0.2042	0.1741
TVDP3	0.8209	0.0699	0.0068	0.1299	0.0306
UK1	0.5483	0.0324	0.0037	0.0623	0.0070
ZC1	0.7517	0.2168	0.1956	0.2700	0.2101
ZC2	0.2079	0.0434	0.0434	0.0434	0.0434
ZC3	0.7517	0.2168	0.1956	0.2700	0.2101
ZC4	0.7517	0.2168	0.1956	0.2700	0.2101
ZC5	0.7517	0.2168	0.1956	0.2700	0.2101
ZY1	0.5104	0.1670	0.0874	0.2164	0.1177
TL1	0.8849	0.2129	0.1129	0.2556	0.1498

Notes on metrics:

fb_F=frame-based F-measure

eb_F_500ms=onset-only event-based F-measure (500ms tolerance)

eb_Foff_500ms=onset-offset event-based F-measure (500ms tolerance)

eb_F_1s=onset-only event-based F-measure (1s tolerance)

eb_Foff_1s=onset-offset event-based F-measure (1s tolerance)

Individual Results Files for Task 2

JS2= Jan Schlüter
MM3= Matija Marolt
TVDP1= Nikolaos Tsipas, Lazaros Vrysis, Charalampos Dimoulas, George Papanikolaou
TVDP2= Nikolaos Tsipas, Lazaros Vrysis, Charalampos Dimoulas, George Papanikolaou
TVDP3= Nikolaos Tsipas, Lazaros Vrysis, Charalampos Dimoulas, George Papanikolaou
UK1= Aiko Uemura, Jiro Katto
ZC1= Chaogang Zhang, Chuan-Yi Chen
ZC2= Chaogang Zhang, Chuan-Yi Chen
ZC3= Chaogang Zhang, Chuan-Yi Chen
ZC4= Chaogang Zhang, Chuan-Yi Chen
ZC5= Chaogang Zhang, Chuan-Yi Chen
ZY1= Wei-Qiang Zhang, Xu-Kui Yang
TL1= Thomas Lidy (TBA)

Note: detailed results are stored as binary Matlab files (.mat)