Audio classification from time-frequency texture

Yu, Guoshen; Slotine, Jean‐Jacques E.

doi:10.1109/icassp.2009.4959924

Cited by 64 publications

(41 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, in [17], a texture-of-texture approach is used, which recursively applies filters to spectrogram and MFCC images to capture certain shapes, such as horizontal and vertical lines in the image, can classify music into one of three genres. Another approach in [19] uses local image patches, which are randomly selected during training, to classify musical instruments. A feature is then extracted for classification based on the minimum matching energy of each patch over the testing spectrogram.…”

Section: Related Work a Literature Reviewmentioning

confidence: 99%

Image Feature Representation of the Subband Power Distribution for Robust Sound Event Classification

Dennis

Tran

Chng

2013

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

The ability to automatically recognize a wide range of sound events in real-world conditions is an important part of applications such as acoustic surveillance and machine hearing. Our approach takes inspiration from both audio and image processing fields, and is based on transforming the sound into a two-dimensional representation, then extracting an image feature for classification. This provided the motivation for our previous work on the spectrogram image feature (SIF). In this paper, we propose a novel method to improve the sound event classification performance in severe mismatched noise conditions. This is based on the subband power distribution (SPD) image-a novel two-dimensional representation that characterizes the spectral power distribution over time in each frequency subband. Here, the high-powered reliable elements of the spectrogram are transformed to a localized region of the SPD, hence can be easily separated from the noise. We then extract an image feature from the SPD, using the same approach as for the SIF, and develop a novel missing feature classification approach based on a nearest neighbor classifier ( NN). We carry out comprehensive experiments on a database of 50 environmental sound classes over a range of challenging noise conditions. The results demonstrate that the SPD-IF is both discriminative over the broad range of sound classes, and robust in severe non-stationary noise.Index Terms-Sound event classification, subband power distribution (SPD), spectrogram, missing feature theory.

show abstract

Section: Related Work a Literature Reviewmentioning

confidence: 99%

Image Feature Representation of the Subband Power Distribution for Robust Sound Event Classification

Dennis

Tran

Chng

2013

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

show abstract

“…The Short-Time Fourier Transform (STFT) was used to calculate the spectrogram ‫ݏ‬ሺ‫ݔ‬ǡ ‫ݕ‬ሻ [12], and the frames were taken to be 256-point frames with 192-point overlap.…”

Section: B Environmental Sound Spectrogram Calculationmentioning

confidence: 99%

Environmental sound classification using log-Gabor filter

Souli

Lachiri

2012

2012 IEEE 11th International Conference on Signal Processing

View full text Add to dashboard Cite

This paper presents novel approaches for efficient feature extraction using environmental sound magnitude spectrogram. We propose approaches based on the visual domain, the spectrogram is passed through a bank of 12 logGabor filters, followed by an averaged operation and passed through an optimal feature selection procedure based on mutual information. The proposed methods were tested on a database of 10 sound classes. The evaluation system is realized by using the multiclass support vector machines (SVM's) that gave rise to a recognition rate of the order 89.62 %.

show abstract

“…The time-frequency representation (TFR) contains the complete information of an audio signal in both spectral and temporal domain [7]. The digital audio stream is divided into segments of fixed length for spectrogram generation using TFR.…”

Section: Time-frequency Representationmentioning

confidence: 99%

“…Little attentions have been given to audio classification in the visual domain. The time-frequency based (spectrogram) features can be used in audio discrimination in the analogous way of image classification [6] [7].…”

Section: Introductionmentioning

confidence: 99%

Spectrogram based features selection using multiple kernel learning for speech/music discrimination

Nilufar

Ray

Molla

et al. 2012

2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

This paper presents a multiple kernel learning (MKL) approach to speech/music discrimination (SMD). The timefrequency representation (spectrogram) implemented by short-time Fourier transform (STFT) of audio segment is decomposed by wavelet packet transform into different subband levels. The subbands, which contain rich texture information, are used as features for this discrimination problem. MKL technique is used to select the optimal subbands to discriminate the audio signals. The proposed MKL based algorithm is applied for SMD of a standard dataset. The experimental results show that the proposed technique yields noticeable improvements in classification accuracy and tolerance toward different noise types compared to the existing methods.

show abstract

Audio classification from time-frequency texture

Cited by 64 publications

References 13 publications

Image Feature Representation of the Subband Power Distribution for Robust Sound Event Classification

Image Feature Representation of the Subband Power Distribution for Robust Sound Event Classification

Environmental sound classification using log-Gabor filter

Spectrogram based features selection using multiple kernel learning for speech/music discrimination

Contact Info

Product

Resources

About