提出一种基于用户语音情感分析的景区反馈评估方法.首先,构建一个面向景区评估的情感分析框架,采用梅尔频率倒谱系数(Mel Frequency Cepstral Coefficients,MFCC)提取语音特征.其次,利用长短期记忆(Long Short-Term Memory,LSTM)网络对提取的特征进行情感极性分类,将情感分为积极、消极、中性.最后,在交互式情感二元动作捕捉(Interactive Emotional Dyadic Motion Capture,IEMOCAP)数据集上进行实验.实验结果显示,本方法在精确率、召回率、准确率等指标上均表现出色,特别是在中性情感分类中达到了较高的识别性能.
Scenic Spot Feedback Evaluation Method Based on User Voice Sentiment Analysis
A feedback evaluation method of scenic spots based on user voice sentiment analysis is proposed.Firstly,a sentiment analysis framework for scenic spot evaluation is constructed,which uses Mel Frequency Cepstral Coefficients(MFCC)to extract speech features.Secondly,the Long Short-Term Memory(LSTM)network is used to classify the extracted features into emotional polarity,and emotions are divided into positive,negative and neutral.Finally,experiments are carried out on Interactive Emotional Dyadic Motion Capture(IEMOCAP)data sets.The experimental results show that this method performs well in accuracy,recall and accuracy,especially in neutral emotion classification.
voice sentiment analysisscenic spot evaluationMel Frequency Cepstral Coefficients(MFCC)Long Short-Term Memory(LSTM)network
NETL
NSTL
万方数据
