2015:Audio Fingerprinting Results

From MIREX Wiki
Revision as of 23:49, 13 July 2016 by Chung-Che Wang (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Introduction

These are the results for the 2015 running of the Audio Fingerprinting task. For background information about this task set please refer to the 2015:Audio Fingerprinting page.

General Legend

Sub code Submission name Abstract Contributors
CZ3 CYCG3 PDF ChuanYi Chen, Chaogang Zhang
CZ4 CYCG4 PDF ChuanYi Chen, Chaogang Zhang
CZ5 CYCG5 PDF ChuanYi Chen, Chaogang Zhang
SW3 ACRCloud_1 PDF Steve Wang
SW4 ACRCloud_2 PDF Steve Wang
YCP4 Sogou_AFP_V1_20150914 PDF Guangchao Yao,Yiqian Pan, Wei Chen
YCP5 Sogou_AFP_V2_20150914 PDF Guangchao Yao,Yiqian Pan, Wei Chen
YCP6 Sogou_AFP_V3_20150914 PDF Guangchao Yao,Yiqian Pan, Wei Chen
ZW1 fingerprint_Mask1 PDF Zhichao Wang
ZW2 fingerprint_MASK2 PDF Zhichao Wang

Summary Results

Size of database and running time are announced together with top-1 hit rate in the following table. Note that running time is announced for reference because machine spec, machine status, and number of cores used by a submission are different. IDs and specs of machines are:

  • A and B: 1.9 GHz, 24 cores CPU, 64 GB RAM
  • C: a VM with 8 cores CPU and 32 GB RAM, hosted on 2.4 GHz, 32 cores CPU and 128 GB RAM
  • D: 2.1 GHz, 24 cores CPU, 32 GB RAM


Sub
code
Time/machine for builder
(hh:mm:ss/machine ID)
Size of database
(GB)
Time/machine for matcher
(hh:mm:ss/machine ID)
Top-1 hit rate
(%)
CZ3 06:32:06 (C) 560 00:32:08 (C) 88.77
CZ4 06:20:51 (C) 560 00:31:48 (C) 88.77
CZ5 06:20:05 (C) 1156 00:33:14 (B) 90.78
SW3 20:03:35 (A) 1513 02:47:30 (B) 91.85
SW4 20:59:08 (A) 2212 01:49:50 (C) 91.88
YCP4 00:35:26 (A) 2012 00:55:35 (A) 84.72
YCP5 00:43:19 (A) 2121 00:57:06 (A) 85.17
YCP6 00:45:03 (A) 2176 00:57:18 (A) 85.28
ZW1 05:04:38 (C) 1786 00:36:59 (C) 65.18
ZW2 05:30:50 (A) 1438 01:05:02 (A) 68.11

download these results as csv

  • Correction: "YCP" should be "YPC", "GB" should be "MB", all "C" should be "D"