基于遗传算法提高计算机连续语音识别率的研究

为提高计算机连续语音识别准确率,应用遗传算法人工智能控制技术具有的高效全局启发式搜索特点,探讨了提高计算机连续语音识别准确率的优化方法,通过提取语音特征参数,作为初始语音群体的生成,把语音识别内核程序中语言焦点作为遗传算法中操作算子,以高效全局启发式适配语音类型,连续语音识别准确率提高7％,为实现真正意义上计算机“人一机对话”进行了大胆的尝试。

Research on continuing speech identifying by computer on genetic algorithm

With the efficient and overall enlighten searching characteristics of Genetic Algorithm which is the artificial intelligence controlling technique, the method of improving the computer's ability in identifying the continuing speech is studied. The computer's speech characteristics is used as the original speeches produced,the focus speech in speech identified inner program is used as the operating factor of Genetic Algorithm to identify the continuing speech , the speech types is adapted as efficient and overall enlighten . The ability to identify the continuing speech is improved by 7%.In this process, it proceeds adventruous trials to search the true " person to machine dialogue" mechanism.