Low-power audio classification for ubiquitous sensor networks

Ravindran, Sourabh; Anderson, David V.; Slaney, Malcolm

doi:10.1109/icassp.2004.1326832

Cited by 21 publications

(11 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We refer to this representation as noise-robust auditory features (NRAF). The noise robustness of these features is shown elsewhere [2]. The NRAF extraction can be implemented in low-power analog VLSI circuitry as shown in Figure 5.…”

Section: Noise-robust Auditory Featuresmentioning

confidence: 99%

See 1 more Smart Citation

A Physiologically Inspired Method for Audio Classification

Ravindran

Schlemmer

Anderson

2005

EURASIP J. Adv. Signal Process.

View full text Add to dashboard Cite

We explore the use of physiologically inspired auditory features with both physiologically motivated and statistical audio classification methods. We use features derived from a biophysically defensible model of the early auditory system for audio classification using a neural network classifier. We also use a Gaussian-mixture-model (GMM)-based classifier for the purpose of comparison and show that the neural-network-based approach works better. Further, we use features from a more advanced model of the auditory system and show that the features extracted from this model of the primary auditory cortex perform better than the features from the early auditory stage. The features give good classification performance with only one-second data segments used for training and testing

show abstract

Section: Noise-robust Auditory Featuresmentioning

confidence: 99%

“…We present features extracted from a model of the early auditory system that have been shown to be robust to noise [1,2]. The feature extraction can be implemented in lowpower analog VLSI circuitry which apart from providing substantial power gains also enables us to achieve feature extraction in real time.…”

Section: Introductionmentioning

confidence: 99%

A Physiologically Inspired Method for Audio Classification

Ravindran

Schlemmer

Anderson

2005

EURASIP J. Adv. Signal Process.

View full text Add to dashboard Cite

show abstract

“…One of the most popular features currently used in many fields of audio research, are the mel-frequency cepstral coefficients (MFCCs). MFCCs have been shown to outperform other methods in a number of different research areas [19]. They are also appealing because their computational cost is low.…”

Section: Feature Extractionmentioning

confidence: 99%

Determining the best proportion of music genre to be played in a radio program

Keshtkar

Bastanfard

2015

2015 7th Conference on Information and Knowledge Technology (IKT)

View full text Add to dashboard Cite

In this paper a scheme for a monitoring system is proposed that can determine the best proportion of music genre to be played in a radio program based on music genre variety and audience feedback. An audio fingerprinting algorithm has been used to determine the music genres played through a radio program. This method is noise and distortion resistant and very scalable. Moreover a feedback system has been employed to gather feedback from audience. Finally an evaluation is presented by observing the data from both steps. To find the proper proportion, the proportion of music genres was changed each time and monitored the system thoroughly to see whether the results change or not. In this approach, the audience behavior can be monitored in music listening and the best proportion of music genre will be determined with a quite good accuracy as well.

show abstract

“…A classification algorithm trained using clean test sequences may fail to work properly when the actual testing sequences contain background noise with certain SNR levels (see test results in [5] and [6]). The so-called early auditory model proposed by Wang and Shamma [7] has proved to be robust in noisy environments because of an inherent self-normalization property which causes noise suppression.…”

Section: Introductionmentioning

confidence: 99%

A simplified early auditory model with application in audio classification

Wei

Champagne

2006

Can. J. Electr. Comput. Eng.

View full text Add to dashboard Cite

The past decade has seen extensive research on audio classification and segmentation algorithms. However, the effect of background noise on classification performance has not been widely investigated. Recently, an early auditory model that calculates a so-called auditory spectrum has achieved excellent performance in audio classification along with robustness in a noisy environment. Unfortunately, this early auditory model is characterized by high computational requirements and the use of nonlinear processing. In this paper, certain modifications are introduced to develop a simplified version of this model which is linear except for the calculation of the square-root value of the energy. Speech/music and speech/non-speech classification tasks are carried out to evaluate the classification performance, with a support vector machine (SVM) as the classifier. Compared to a conventional fast Fourier transform-based spectrum, both the original auditory spectrum and the proposed simplified auditory spectrum show more robust performance in noisy test cases. Test results also indicate that despite a reduced computational complexity, the performance of the proposed simplified auditory spectrum is close to that of the original auditory spectrum.La dernière décennie a connu une expansion de la recherche sur les algorithmes de classification audio et de segmentation. Cependant, l'effet du bruit de fond sur les performances de la classification n'a pasété largementétudié. Récemment, un modèle auditif qui calcule un spectre auditif a atteint une performance excellente en classification audio ainsi qu'une robustesse dans un environnement bruité. Malheureusement, ce modèle auditif est caractérisé par des besoinsélevés en calcul et par un traitement non-linéaire. Dans ce papier, quelques modifications sont introduites afin de développer une version simplifiée de ce modèle qui est linéaireà l'exception du calcul de la valeur de la racine carrée de l'énergie. Des tâches de classification de la parole/musique de même que de la parole/non-parole sont effectuées pourévaluer la performance de la classification, en utilisant un classifieurà automatè a support vectoriel. Comparéà une transformation rapide de Fourier conventionnelle, les deux spectres auditifs -celui d'origine et celui simplifié proposé -montrent des performances plus robustes dans les tests avec bruit. Les résultats des tests montrentégalement qu'en dépit d'une complexité de calcul réduite, la performance du spectre auditif simplifié qui aété proposé est proche du spectre auditif d'origine.

show abstract

Low-power audio classification for ubiquitous sensor networks

Cited by 21 publications

References 7 publications

A Physiologically Inspired Method for Audio Classification

A Physiologically Inspired Method for Audio Classification

Determining the best proportion of music genre to be played in a radio program

A simplified early auditory model with application in audio classification

Contact Info

Product

Resources

About