基于视听信息决策融合的自动语音识别方法
首发时间:2010-12-31
摘要:以噪声环境下自动语音识别(automatic speech recognition,ASR)研究为背景,建立视听信息决策层融合模型进行联合事件判断提高抗噪声能力,即在隐马尔可夫(HMM)统计模型的基础上,通过融合处理来降低或消除融合中的音频噪声成分。通过HMM的训练步骤估计模型的参数,然后由关联处理进行决策层融合判决,最终获得联合推断结果即基于加权后验概率的融合模型进行融合判决。仿真结果表明,应用视听信息融合能克服音频噪声,提高识别准确率。
关键词: 自动语音识别 视听信息决策融合 隐马尔可夫 决策层融合
For information in English, please click here
Automatic Speech Recognition Metheod Based on Audio_visual Information Decion Fusion
Abstract:With automatic speech recognition of noisy environment research background,audio-visual information decision fusion model is established to joint event judge for increasing anti-noise ability,which reduce or eliminate the audio noise components with fusion on the basis of statistical hidden markov madel(HMM).This model makes decision by related process and finally obtains joint judge results based on weighting posteriori probability model fusion through the training estimation of HMM. Simulation results show that the application of audio-visual information decision fusion can overcome the audio visual noise and improve identification accuracy.
Keywords: automata speech recognition audio-visual information decision fusion HMM decision fusion
基金:
论文图表:
引用
No.4401027561222129****
同行评议
共计0人参与
勘误表
基于视听信息决策融合的自动语音识别方法
评论
全部评论0/1000