Direct Recovery of Clean Speech Using a Hybrid Noise Suppression Algorithm for Robust Speech Recognition System

Table 1

Detailed recognition rates (%).

Iteration No.	1	2	3	4	5
Clean	99.09	99.08	98.37	98.36	97.23
Avg 0–20	85.70	85.76	85.66	80.07	76.17
−5 dB	27.80	27.85	23.24	23.24	21.31