Table 2 Recognition rate (%) obtained by proposed and standard methods with suburban train noise
From: A bio-inspired feature extraction for robust speech recognition
Recognition rate with HMM-4-GM | ||||||
|---|---|---|---|---|---|---|
SNR level | PLPaGc | PLP | LPCC | LPC | MFCC | |
0 dB | 38.55 | 27.77 | 21.79 | 11.86 | 26.95 | |
5 dB | 65.59 | 50.16 | 40.48 | 13.62 | 49.42 | |
Suburban train noise | 10 dB | 84.71 | 72.74 | 60.96 | 18.47 | 71.66 |
15 dB | 92.74 | 85.82 | 77.90 | 28.96 | 86.30 | |
20 dB | 95.77 | 91.72 | 87.06 | 41.96 | 92.60 | |
Average | 75.47 | 65.64 | 57.64 | 22.97 | 65.39 | |