Intra-frame cepstral sub-band weighting and histogram equalization for noise-robust speech recognition

EURASIP Journal on Audio, Speech, and Music Processing

Table 4 The recognition accuracy results (%) of the MFCC baseline, CHN, and eight forms of WS-HEQ

Method		Clean	20 dB	15 dB	10 dB	5 dB	0 dB	−5 dB
MFCC		99.12	95.33	86.62	65.93	36.01	14.86	8.16
CHN		98.97	96.30	93.89	88.48	74.81	48.10	19.94
${WS-HEQ}_{I}^{(1)}$	α=1.0 (S-HEQ)	99.02	97.24	94.99	90.09	77.88	52.73	22.68
	α=0.6	98.99	97.60	95.72	91.54	80.62	55.87	23.26
${WS-HEQ}_{I}^{(2)}$	α=1.0	99.02	97.34	95.14	90.10	77.38	49.57	18.40
	α=0.6	99.09	97.68	95.64	91.30	79.23	49.97	17.69
${WS-HEQ}_{I}^{(3)}$	α=1.0	98.96	96.89	94.45	89.02	75.55	47.78	17.26
	α=0.5	98.99	97.51	95.71	91.59	80.59	55.95	23.88
${WS-HEQ}_{I}^{(4)}$	α=1.0	98.93	97.20	95.01	89.75	76.57	47.68	16.92
	α=0.7	99.07	97.87	96.18	91.98	79.99	51.94	19.83
${WS-HEQ}_{II}^{(1)}$	α=1.0	98.84	96.86	94.61	89.76	78.00	54.30	24.91
	α=0.6	99.05	97.53	95.75	91.99	81.35	58.35	26.98
${WS-HEQ}_{II}^{(2)}$	α=1.0	98.87	97.41	95.50	91.15	79.06	53.96	22.29
	α=0.6	99.04	97.60	95.70	91.68	80.83	56.61	24.45
${WS-HEQ}_{II}^{(3)}$	α=1.0	98.89	96.76	94.29	88.97	76.42	52.44	23.69
	α=0.7	99.05	97.87	95.98	91.86	80.62	53.88	20.51
${WS-HEQ}_{II}^{(4)}$	α=1.0	98.92	97.13	94.84	90.07	76.92	49.96	19.89
	α=0.6	98.96	97.49	95.54	91.17	79.00	51.23	19.37

They are for different SNR cases while averaged over ten noise situations as to the Aurora-2 database.