Table 12 The retrieval performances obtained with the class vectors of an MLP classifier
From: An evolutionary feature synthesis approach for content-based audio retrieval
FS | Basic database (6 classes) | Extended database (12 classes) | ||
|---|---|---|---|---|
ANMRR | AP (%) | ANMRR | AP (%) | |
Segment | ||||
STAT | 0.334 (1) | 66.0 (1) | 0.620 (2) | 37.1 (2) |
MFCC | 0.089 (1) | 90.4 (1) | 0.532 (1) | 45.7 (1) |
Δ-MFCC | 0.340 (2) | 65.3 (2) | 0.680 (1) | 30.9 (1) |
ΔΔ-MFCC | 0.320 (1) | 66.8 (1) | 0.634 (2) | 35.3 (2) |
LPC | 0.489 (1) | 50.2 (1) | 0.704 (2) | 28.7 (2) |
LPCC | 0.562 (1) | 42.6 (1) | 0.746 (1) | 24.4 (1) |
S_AUDIO | 0.255 (1) | 73.6 (1) | 0.561 (2) | 43.2 (2) |
Key-frame | ||||
MFCC + deltas | 0.334 (1) | 64.8 (1) | 0.555 (1) | 42.7 (1) |
LPC + LPCC | 0.669 (1) | 32.1 (1) | 0.805 (1) | 18.8 (1) |
K_AUDIO | 0.391 (2) | 59.4 (2) | 0.604 (1) | 38.0 (1) |