Skip to main content

Table 9 The retrieval performance statistics obtained using the EFS technique with fixed output dimension and only two operators

From: An evolutionary feature synthesis approach for content-based audio retrieval

FS

Basic database (6 classes)

Extended database (12 classes)

ANMRR (μ ± σ)

AP (%) (μ ± σ)

ANMRR (μ ± σ)

AP (%) (μ ± σ)

Segment

    

STAT

0.278 ± 0.016

70.4 ± 1.7

0.534 ± 0.024

44.1 ± 2.4

MFCC

0.280 ± 0.022

69.6 ± 2.1

0.528 ± 0.023

44.9 ± 2.2

Δ-MFCC

0.375 ± 0.016

60.6 ± 1.6

0.574 ± 0.011

40.6 ± 1.0

ΔΔ-MFCC

0.393 ± 0.013

59.0 ± 1.3

0.598 ± 0.020

38.3 ± 2.0

LPC

0.549 ± 0.006

43.8 ± 0.6

0.729 ± 0.013

25.7 ± 1.2

LPCC

0.589 ± 0.015

39.6 ± 1.5

0.740 ± 0.006

24.8 ± 0.5

S_AUDIO

0.236 ± 0.015

74.7 ± 1.6

0.480 ± 0.013

49.9 ± 1.2

Key-frame

    

MFCC + deltas

0.463 ± 0.024

51.4 ± 2.3

0.629 ± 0.030

35.2 ± 2.8

LPC + LPCC

0.709 ± 0.007

28.4 ± 0.7

0.813 ± 0.003

18.0 ± 0.3

K_AUDIO

0.346 ± 0.016

63.0 ± 1.5

0.532 ± 0.012

44.4 ± 1.2