车间环境下机器人语音控制的特征提取算法

doi:10.19665/j.issn1001-2400.2020.02.003

摘要/Abstract

摘要：

针对移动机器人在车间实际工作环境中,由于噪声的影响导致对语音控制命令识别性能差的问题,提出了一种基于伽玛通倒谱系数和Teager能量算子混合特征提取的新算法。该算法用伽玛通滤波器代替抗噪性比较普通的梅尔滤波器,在提取伽玛通倒谱系数的过程中加入反映语音信号能量的Teager能量算子组成新的特征,并考虑语音信号的动态特性,将其与一阶差分参数融合组成混合特征;应用主成分分析法降维,将得到的混合特征用于移动机器人控制命令的语音识别系统。实验结果表明,在车间噪声以及信噪比为10dB的环境下,混合特征的识别率较梅尔倒谱系数提高了12.20%,通过主成分分析法得到的混合特征的识别率提高了1.02%。

关键词: 伽玛通滤波器, Teager能量算子, 特征提取, 机器人控制

Abstract:

In the real working environment,the mobile robots have a poor recognition performance to speech control commands due to the noise effect. Aiming at this issue,this paper proposes a new algorithm based on the gammatone frequency cepstral coefficient and the mixed feature extraction of the Teager energy operator. This algorithm replaces the common Mel filter with the Gammatone filter. In the process of extracting gammatone frequency cepstral coefficients,the Teager energy operator reflecting the energy of speech signal is added to form a new feature, with the dynamic characteristics of the speech signal considered. It is combined with the first-order difference parameters to form a mixed feature. And the principal component analysis is made to reduce the dimension,and the final mixed features are used to the speech recognition system for control command of the mobile robot. Experimental results show that,in the environment of the workshop noise and signal-to-noise ratio of 10dB,the recognition rate of mixed features is improved by 12.20% compared with the mel frequency cepstrum coefficient. The recognition rate of the mixed feature is increased by 1.02% when the dimension is reduced by principal component analysis.

Key words: Gammatone filter, Teager energy operator, feature extraction, robot control

中图分类号:

TN912.34

王晓华,要鹏超,马丽萍,王文杰,张蕾. 车间环境下机器人语音控制的特征提取算法[J]. 西安电子科技大学学报, 2020, 47(2): 16-22.

WANG Xiaohua,YAO Pengchao,MA Liping,WANG Wenjie,ZHANG Lei. Algorithm for extraction of features of robot speech control in the factory environment[J]. Journal of Xidian University, 2020, 47(2): 16-22.

图/表 9

图1

图2

图3

图4

图5

图6

表1

表2

图7

参考文献 15

[1]	ISHI C T, ISHIGURO H, HAGITA N . Analysis of Head Motions and Speech, and Head Motion Control in an Android Robot[C]// Geminoid Studies: Science and Technologies for Humanlike Teleoperated Androids. Singapore: Springer Singapore, 2018: 89-110.
[2]	PIRES J N, AZAR A S . Advances in Robotics for Additive/Hybrid Manufacturing: Robot Control, Speech Interface and Path Planning[J]. Industrial Robot, 2018,45(3):311-327.
[3]	MAGASSOUBA A, BERTIN N, CHAUMETTE F . Exploiting the Distance Information of the Interaural Level Difference for Binaural Robot Motion Control[J]. IEEE Robotics and Automation Letters, 2018,3(3):2048-2055.
[4]	KUMAR A, ROUT S S, GOEL V . Speech Mel Frequency Cepstral Coefficient Feature Classification Using Multi Level Support Vector Machine[C]// Proceedings of the 2017 4th IEEE Uttar Pradesh Section International Conference on Electrical, Computer and Electronics. Piscataway: IEEE, 2017: 134-138.
[5]	SHAHIN I . Emotion Recognition Based on Third-order Circular Suprasegmental Hidden Markov Model[C]// Proceedings of the 2019 IEEE Jordan International Joint Conference on Electrical Engineering and Information Technology. Piscataway: IEEE, 2019: 800-805.
[6]	ISHIMITSU S . Speech Recognition Method and Speech Recognition Apparatus: US6067513A[P]. 2000 -05-23.
[7]	TAZI E B . A Robust Speaker Identification System Based on the Combination of GFCC and MFCC Methods[C]// Proceedings of the 2017 International Conference on Multimedia Computing and Systems. Washington: IEEE Computer Society, 2017: 54-58.
[8]	CAN G, AKBAS C E, CETIN A E . Recognition of Vessel Acoustic Signatures Using Non-linear Teager Energy Based Features[C]// Proceedings of the 2016 International Workshop on Computational Intelligence for Multimedia Understanding. Piscataway: IEEE, 2016: 7801190.
[9]	白静, 史燕燕, 薛珮芸 , 等. 融合非线性幂函数和谱减法的CFCC特征提取[J]. 西安电子科技大学学报, 2019,46(1):86-92.
	BAI Jing, SHI Yanyan, XUE Peiyun , et al. CFCC Feature Extraction for Fusion of the Power-law Nonlinearity Function and Spectral Subtraction[J]. Journal of Xidian University, 2019,46(1):86-92.
[10]	MARKOVIĆ B, GALIĆ J, GROZDIĆ D , et al. Whispered Speech Recognition Based on Gammatone Filterbank Cepstral Coefficients[J]. Journal of Communications Technology and Electronics, 2017,62(11):1255-1261.
[11]	BANDELA S R, KUMAR T K . Stressed Speech Emotion Recognition Using Feature Fusion of Teager Energy Operator and MFCC[C]// Proceedings of the 2017 8th International Conference on Computing, Communications and Networking Technologies. Piscataway: IEEE, 2017: 8204149.
[12]	MONTEIRO R L D C, PEREIRA V, COSTA H G . Analysis of the Better Life Index trough a Cluster Algorithm[J]. Social Indicators Research, 2019,142(2):477-506.
[13]	TONG Z, SHI D, YANG S . Scene SLAM: a SLAM Framework Combined with Scene Detection[C]// Proceedings of the 2017 IEEE International Conference on Robotics and Biomimetics. Piscataway: IEEE, 2017: 1-8.
[14]	KORSHUNOV P, GONÇALVES A R, VIOLATO R P V , et al. On the Use of Convolutional Neural Networks for Speech Presentation Attack Detection[C]// Proceedings of the 2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis. Piscataway: IEEE, 2018: 1-8.
[15]	张雪英, 牛溥华, 高帆 . 基于DNN-LSTM的VAD算法[J]. 清华大学学报(自然科学版), 2018,58(5):509-515.
	ZHANG Xueying, NIU Puhua, GAO Fan . VAD Algorithm Based on DNN-LSTM[J]. Journal of Tsinghua University (Science and Technology), 2018,58(5):509-515.

特征参数	识别率/%
MFCC	95.42
GFCC	93.16
TEOGFCC	94.24
ΔTEOGFCC+TEOGFCC	95.33

噪声类型	特征提取算法	信噪比/dB
噪声类型	特征提取算法	-5	0	5	10	15	20
白噪声	MFCC	35.25	41.24	62.05	71.71	80.24	85.25
	GFCC	40.25	48.35	64.25	76.25	85.65	87.75
	TEOGFCC	48.81	65.21	74.24	80.20	87.21	90.20
	TEOGFCC+ΔTEOGFCC	50.64	67.32	77.58	84.38	89.01	91.14
粉红噪声	MFCC	36.70	42.13	63.08	72.31	81.25	86.69
	GFCC	41.25	49.05	66.21	75.65	84.36	88.82
	TEOGFCC	49.25	66.32	74.06	80.35	88.31	90.05
	TEOGFCC+ΔTEOGFCC	51.88	68.24	77.86	85.63	90.32	91.78
汽车噪声	MFCC	34.24	42.25	63.58	71.12	81.33	86.96
	GFCC	43.21	50.26	68.32	76.30	85.65	87.24
	TEOGFCC	47.65	63.45	73.26	80.09	87.87	90.74
	TEOGFCC+ΔTEOGFCC	50.24	67.69	78.35	84.25	89.81	91.65
车间噪声	MFCC	35.58	43.28	65.36	73.25	82.16	85.23
	GFCC	42.32	51.25	67.48	75.91	86.51	87.24
	TEOGFCC	48.04	61.68	72.55	81.48	86.61	91.34
	TEOGFCC+ΔTEOGFCC	51.64	68.28	79.85	85.45	90.81	92.08