基于双微麦克风阵列与Wide ResNet网络的语音命令词识别
SPEECH COMMAND WORD RECOGNITION BASED ON DUAL MICRO MICROPHONE ARRAY AND WIDE RESNET
祁潇潇 1曾庆宁 1赵学军1
作者信息
- 1. 桂林电子科技大学信息与通信学院 广西桂林 541004
- 折叠
摘要
为了提高噪声环境下语音识别的稳健性[1],提出宽残差深度神经网络的语音识别算法.该算法结合双微麦克风阵列系统、语音数据集为双微麦克风数据集,使用功率归一化倒谱系数作为特征参数输入到残差网络中进行训练.实验表明,与ResNetl5模型、ResNet18模型相比,只有三个残差模块的宽残差网络在噪声环境下语音命令词的识别和内外部说话人检测任务中具有较高的准确度,均达到了 95％以上.
Abstract
In order to improve the robustness of speech recognition in noise environment,a speech recognition algorithm based on wide residual deep neural network is proposed.The algorithm combined the dual micro microphone array system,and the voice data set was the dual micro microphone data set.The power normalized cepstrum coefficient was used as the characteristic parameter to input into the residual network for training.Experimental results show that,compared with the Resnet15 model and Resnet18 model,the wide ResNet with only three residual modules has higher accuracy in the recognition of speech command words and the internal and external speaker detection task under noise environment,both reaching more than 95％.
关键词
语音识别/宽残差神经网络/功率归一化倒谱系数/双微麦克风阵列Key words
Speech recognition/Wide ResNet/Power normalized cepstrum coefficient/Dual micro microphone array引用本文复制引用
基金项目
国家自然科学基金(61961009)
广西自然科学基金重点项目(2016GXNSFDA380018)
广西无线宽带通信与信号处理重点实验室项目(GXKL06200107)
出版年
2024
国家科技期刊平台
NSTL
万方数据