Spectrogram Image Feature for Sound Event Classification in Mismatched Conditions

Acoustic feature extraction by statistics based local binary pattern for environmental sound classification

“…The performance results are shown in Table 2b. Some other works have also reported the classification performance on the RWCP dataset, but they used the subset of the dataset; e.g., the works of [5,6,17,18] reported around 90% only on 10 ∼ 20 sound categories, and Dennis et al [4] exhibited the performance of 98.1% which is close to our result though their method was evaluated only on 60 categories, half subset of ours. Therefore, we can say that the proposed method achieves the state-of-the-art performance on the whole RWCP dataset.…”

“…While the methods for classifying speech and music have been intensively developed for decades, those for the environmental sounds are studied with keen attention in recent years [1,2,3,4]. The environmental sounds are different from the speech and music in that the acoustic signals are not stationary nor well-structured; characteristics in these types of sounds are discussed in [5].…”

“…Guo et al [6] extract the time-frequency intersection patterns from the spectrogram similarly to image projection profile. Dennis et al [4] establish the moment-based spectrogram features.…”

Exploring Time Series Spectral Features in Viral Hashtags Prediction

“…Scalograms are considered textures and 2D DWT have been used to analyze these textures in many applications [14] [15]. We follow [16] to decompose a scalogram into 2D DWT image.…”

Environmental Sound Recognition Using Spectrogram Image Features

“…Mel-Frequency Cepstral Coefficients (MFCCs), Zero Crossing Rate (ZCR), Spectral Centroid are some of the most widely used features for audio analysis. Many previous efforts have also been made in classifying audio signals using other features such as MPEG-7 descriptors [1,2] , Linear Prediction coefficients [3] , features derived from statistics of spectrogram image of an audio [4] and Log-Gabor Filters [5] . The bag of phrases approach is introduced in [6] , where a codebook is generated using Gaussian Mixture Model and then the codebook is used to obtain a new set of features for the classification.…”