摘要
改进相关主题模型(correlated topic model,CTM)使其具有动态性,提出了动态相关主题模型(dynamic correla-ted topic model,DCTM),使用变分卡尔曼滤波推断模型的隐含主题参数。将DCTM作为降维模型与隐马尔克夫模型(hidden Markov model,HMM)相结合对音乐分类。这一方法将音乐片段分割为等长的小片段,将小片段的声学特征向量通过相似性比较转化为单词序列,通过DCTM将单词序列转换为主题向量。将主题向量输入HMM得出分类结果。由于DCTM的动态建模,更好地提取对分类有用的信息,因此增强了方法的分类能力。实验验证了方法的有效性。
A new model called dynamic correlated topic model （DCTM） introduced, which improved correlated topic model （CTM） and made it has dynamic. A variational Kalman filtering is used to solve the problem of latent topics parameters. A new music classification method combining dynamic correlated topic model （17）CTM） and hidden Markov model （HMM） is proposed. First the acoustic feature vectors of a music clip are transformed into a sequence of words through replacing each feature vector with its most similar word in the vocabulary. Then, the sequence of words is ted into DCTM model to infer a low-dimensional topic vector of the music clip. Finally, HMM model is utilized to calculate the likelihood of the topic vector, acting as an observation sequence of the model, so that classification of the music clip is reached. Because of the DCTM can model the music clip with dynamic, extract more useful information of the classification. So strengthen the ability of the classification method. The results of experiments proves the effectiveness of the method.
出处
《计算机工程与设计》
CSCD
北大核心
2012年第11期4245-4249,4332,共6页
Computer Engineering and Design
关键词
音乐分类
相关主题模型
动态相关主题模型
变分卡尔曼滤波
隐马尔克夫模型
music classification
correlated topic model
dynamic correlated topic model
variational Kalman filtering
hidden Markov model
