Research Article
Comparative Evaluation of Single-Channel MMSE-Based Noise Reduction Schemes for Speech Recognition
Table 4
Results of frequency-domain algorithms on AURORA4.
| | Set 01 | Set 02–07 | Set 08 | Set 09–14 | Average |
| | C | M | C | M | C | M | C | M | C | M |
| LSA | 87.92 | 74.48 | 52.07 | 64.48 | 77.08 | 74.99 | 39.00 | 55.15 | 50.82 | 62.16 | | 88.73 | 77.78 | 60.71 | 72.19 | 81.36 | 78.20 | 49.17 | 65.48 | 59.24 | 70.14 | | 88.14 | 77.69 | 54.11 | 65.38 | 76.88 | 74.22 | 40.73 | 57.51 | 52.43 | 63.52 | OMLSA | 88.32 | 79.08 | 56.31 | 69.27 | 77.20 | 76.24 | 43.60 | 61.59 | 54.64 | 67.18 | | 88.99 | 79.74 | 61.04 | 72.66 | 79.77 | 78.82 | 49.06 | 65.93 | 59.24 | 70.72 | | 88.62 | 79.12 | 57.60 | 70.52 | 75.42 | 75.47 | 44.67 | 63.22 | 55.55 | 68.36 |
| Baseline | 89.80 | 84.53 | 54.57 | 74.57 | 77.05 | 81.62 | 42.82 | 67.68 | 53.65 | 72.83 |
|
|