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

EURASIP Journal on Audio, Speech, and Music Processing

Table 6 The recognition accuracy results (%) achieved by the combination of TSN and WS-HEQ

Method	Set A	Set B	Set C	Average
TSN	80.86	82.39	80.46	81.39
${WS-HEQ}_{I}^{(1)}$	83.36	85.37	83.89	84.27
${WS-HEQ}_{I}^{(1)}$ +TSN	85.32	86.77	85.01	85.84
${WS-HEQ}_{I}^{(2)}$	82.29	83.22	82.82	82.76
${WS-HEQ}_{I}^{(2)}$ +TSN	85.06	86.21	84.75	85.46
${WS-HEQ}_{I}^{(3)}$	83.57	85.15	83.93	84.27
${WS-HEQ}_{I}^{(3)}$ +TSN	85.35	86.54	85.15	85.78
${WS-HEQ}_{I}^{(4)}$	82.88	84.70	82.78	83.59
${WS-HEQ}_{I}^{(4)}$ +TSN	84.41	85.81	83.96	84.88
${WS-HEQ}_{II}^{(1)}$	84.13	86.16	84.39	84.99
${WS-HEQ}_{II}^{(1)}$ +TSN	85.59	87.41	85.26	86.25
${WS-HEQ}_{II}^{(2)}$	83.54	85.75	83.83	84.48
${WS-HEQ}_{II}^{(2)}$ +TSN	85.77	87.08	85.34	86.21
${WS-HEQ}_{II}^{(3)}$	83.25	85.10	83.50	84.04
${WS-HEQ}_{II}^{(3)}$ +TSN	84.68	86.75	84.25	85.42
${WS-HEQ}_{II}^{(4)}$	82.30	83.45	82.90	82.88
${WS-HEQ}_{II}^{(4)}$ +TSN	84.59	85.32	84.59	84.88

They are for different test sets while averaged over five SNR conditions (20 to 0 dB) as to the Aurora-2 database. The scaling factor α listed in Table 1 is adopted for each WS-HEQ.