一种粒子群算法在语音识别中的应用

Application of Particle Swarm Optimization Algorithm in Speech Recognition

分析粒子群算法的基本原理及影响因素,理解语音识别流程,将粒子群算法运用到语音识别过程中,根据运行效果,提出一种柯西变异和粒子群聚集度混合扰动的方法。一旦粒子群聚集度偏低就说明粒子群多样性欠缺,通过变异当前最优值扰动粒子群收敛,避免粒子群的过早收敛,改善全局遍历性。实验证明,改进后的粒子群算法较好地提高了全局收敛性,提升了语音识别效果。

The basic principle and influencing factors of particle swarm optimization algorithm are analyzed,the speech recognition process is understand,particle swarm optimization algorithm is applied to the speech recognition process,and a mixed disturbance meth-od of Cauchy variation and particle swarm aggregation according to the operation effect is proposed.The low particle swarm aggregation indicates the lack of particle swarm diversity.By mutating the current optimal value to disturb the particle swarm convergence,the pre-mature convergence of particle swarm is avoided and the global ergodic property is improved.Experiments show that the improved parti-cle swarm optimization algorithm improves the global convergence and the effect of speech recognition.