支持向量机语音识别算法在DM6446上的实现

摘要

针对语音识别系统对实时性和便携性的要求,提出一种基于MFCC/SVM在DM6446嵌入式系统开发平台上的实现方法,实现了一个面向非特定人的语音识别系统,将有向无环图多类分类支持向量机算法移植到该平台.并在该平台用DAG方法对非特定人孤立词和连接词进行语音识别,比隐马尔可夫模型有明显优势.通过样本预选取算法对训练样本进行预选取处理,并且应用到嵌入式语音识别系统中,大大降低了训练时间和测试时间.%In order to meet the need of real-time and portable characteristics of the speech recognition system, this paper proposes a method for speaker-independent word speech recognition system which combines MFCC with SVM on DM6446 embedded speech recognition system, and transplants multiclass classification support vector machines of DAG methods on this system. Using DAG methods in speaker-independent isolated word and connected word speech recognition system, the results are better than hidden Markov models. Pre-selecting the training sample through pre-selection algorithm in embedded speech recognition system, the results greatly reduce training time and testing time.