Acoustic scene classification method and system

• Main Authors: WANG BEI, LI YONGHONG, ZHOU YANBING, SUN CHONGWU, LIU MING
• Format: Patent
• Language: Chinese; English
• Published: 09.03.2021
• Subjects: ACOUSTICS; CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; MUSICAL INSTRUMENTS; PHYSICS; SPEECH ANALYSIS OR SYNTHESIS; SPEECH OR AUDIO CODING OR DECODING; SPEECH OR VOICE PROCESSING; SPEECH RECOGNITION

The invention provides an acoustic scene classification method, which comprises the following steps of S1, inputting an acquired environment audio signal into a feature extraction module, and extracting Mel logarithm spectrum features based on harmonic impact, S2, inputting the extracted Mel logarithm spectrum features into an ensemble learning classifier module, and performing acoustic scene classification to obtain an acoustic scene classification result, and S3, inputting an acoustic scene classification result into a comprehensive decision-making module, carrying out comprehensive decision-making, and then making a final classification decision. The invention further provides an acoustic scene classification system. The beneficial effect of the invention is that the method can reliablyand accurately recognize different scenes in the audio. 本发明提供了一种声学场景分类方法，包括以下步骤：S1、将采集到的环境音频信号输入到特征提取模块，进行基于谐波冲击的梅尔对数谱特征的提取；S2、将提取到的梅尔对数谱特征输入到集成学习分类器模块，进行声学场景分类，得到声学场景分类结果；S3、将声学场景分类结果输入到综合决策模块，进行综合决策，然后做出最终的分类决定。本发明还提供了一种声学