摘要
对高噪声背景下的语音特点进行了分析,提出了适应于高噪声背景下孤立词语音识别的全套方法,进行了系统实现。对本方法和系统进行了详细的实验比较和测试,证明系统在80dB左右接近同类无噪声识别系统的精度,最高可以抵抗100dB的噪声。
After analyzing the speech features in noisy environment, an integrated method for isolated word recognition in noisy environment was proposed, and an experimental system was realized. The detail experiments show that the recognition rate of the system at the noise level about 80 dB is nearly equal to that of the system at noiseless condition, and the system can bear the most noise level about 100 dB.
出处
《声学学报》
EI
CSCD
北大核心
1998年第3期257-265,共9页
Acta Acustica