出版社:National Defense University Barbaros Naval Sciences and Engineering Institute Journal of Naval Science and Engineering
摘要:The term Underwater Acoustic Signal Recognition (UASR) is used for identifying the platforms by some techniques from the acoustic sound signals they produce.In this paper,we propose two different schemes for UASR.In both schemes,the feature extraction is performed using Mel-Frequency Cepstral Coefficients and Linear Predictive Coding derived Cepstral Coefficients which have been extensively utilized in speech recognition.In the first scheme,the features extracted frame by frame are used as a sequence in the representation of the whole signal.The classification of that sequence of vectors is then performed by Hidden Markov Models with various topologies.The second scheme represents the frame features using Bag of Acoustic Words approach.In training stage,all the feature vectors extracted from the input signal are first clustered into a set of acoustic words.Each feature vector is then assigned to an acoustic word.After the frequency of each word is calculated in the input signal,the final representation is performed by the co-occurrence list of the acoustic words.
其他摘要:Su altı Akustik Sinyal Tanıma (SAST) terimi,platformları ürettikleri seslerden bazı teknikler kullanarak tanıma işlemi için kullanılmaktadır.Her gemi makine,pervane,tekne yapısı ve mürettebat alışkanlıklarının birleşiminden meydana gelen kendine özgü özel
关键词:Bag Of Acoustic Words;Circular HMM;Underwater Acoustics;Time-shift invariant;Sound Recognition;and MFCC
其他关键词:Akustik Kelimeler Kümesi;Çevrimsel Saklı Markov Model;Su altı Akustiği;Zaman-Kayması Bağımsız;Ses Tanıma ve MFCC