阅读说明：本技术 语音检测的方法和装置 (The method and apparatus of speech detection ) 是由 郭红敬 李国梁 王鑫山 杨柯 朱虎 于 2018-03-26 设计创作，主要内容包括：一种语音检测的方法和装置,该方法包括：确定待处理数据中的第一数据块的N个分组中的每个分组的能量,其中,N为正整数(S110)；根据N个分组的能量,确定初始的候选噪声集合和初始的候选语音集合,其中,初始的候选噪声集合中的分组的最大能量小于初始的候选语音集合中的分组的最小能量(S120)；根据初始的候选噪声集合中的每个分组的能量,确定初始的噪声门限(S130)；根据初始的候选语音集合以及初始的噪声门限,确定第一次迭代处理的候选噪声集合和第一次迭代处理的候选语音集合,其中,第一次迭代处理的候选噪声集合中的分组的能量均小于或等于初始的噪声门限,第一次迭代处理的候选语音集合中的分组的能量均大于初始的噪声门限(S140)。(A kind of method and apparatus of speech detection, this method comprises: determining the energy of each grouping in N number of grouping of the first data block in pending data, wherein N is positive integer (S110)；According to the energy of N number of grouping, initial candidate noise set and initial candidate speech set are determined, wherein the ceiling capacity of the grouping in initial candidate noise set is less than the least energy (S120) of the grouping in initial candidate speech set；According to the energy of each grouping in initial candidate noise set, initial noise gate (S130) is determined；According to initial candidate speech set and initial noise gate, determine the candidate noise set of first time iterative processing and the candidate speech set of first time iterative processing, wherein, the energy of grouping in the candidate noise set of first time iterative processing is respectively less than or equal to initial noise gate, and the energy of the grouping in the candidate speech set of first time iterative processing is all larger than initial noise gate (S140).)