Difference between revisions of "2018:Automatic Lyrics-to-Audio Alignment Results"

From MIREX Wiki
(Hansen's dataset a cappella)
(Hansen's dataset a cappella)
Line 46: Line 46:
 
==Subtask2: English pop songs==
 
==Subtask2: English pop songs==
 
===Hansen's dataset  a cappella===
 
===Hansen's dataset  a cappella===
<csv>2018/ala/HansensDataset/aCappella/results_htk_acap.csv</csv>
+
<csv>2018/ala/hansen/clean/summary_hansen_clean.csv</csv>
  
 
===Hansen's dataset mixed===
 
===Hansen's dataset mixed===

Revision as of 05:37, 4 November 2018

General Legend

Sub code Submission name Abstract Contributors
CW1 HMM-based Alignment subtask1 [PDF] Chung-Che Wang
CW2 HMM-based Alignment monophone acoustic models subtask2 PDF Chung-Che Wang
CW3 HMM-based Alignment bi-phone acoustic models subtask2 PDF Chung-Che Wang
GSLW1 GMM-HMM (SAT) Models subtask2 PDF Chitralekha Gupta, Bidisha Sharma, Haizhou Li, Ye Wang
GSLW2 SAT+DNN Models subtask2 PDF Chitralekha Gupta, Bidisha Sharma, Haizhou Li, Ye Wang
GSLW3 Non-vocal suppression and SAT Models subtask 2 PDF Chitralekha Gupta, Bidisha Sharma, Haizhou Li, Ye Wang
FFJ1 subtask 1 & 2 [PDF] Yang-Bin Dell Fan, Zhe-Cheng Fan, Jyh-Shing Roger Jang

Results

Subtask1: A cappella Mandarin Chinese pop songs

Source-separated

Mixed

Subtask2: English pop songs

Hansen's dataset a cappella

Submission Mean ASE Mean PCS Mean PCETW
CW2 0.35 0.65 0.79
CW3 0.47 0.62 0.81
GSLW1 2.66 0.39 0.48
GSLW2 2.90 0.35 0.40
GSLW3 5.80 0.31 0.31

download these results as csv

Hansen's dataset mixed

Mauch's dataset

Gracenote dataset a cappella

Gracenote dataset mixed