1,721,005 research outputs found
Anomalous Sound Event Detection Based on One-Class Classification Using Variational Autoencoders and Interval Type-2 Fuzzy Sets
No full text
Modeling uncertainty with interval-valued type-2 fuzzy sets: Application to anomalous sound event detection
No full textSince a few years, audio signal processing has been focused on detecting audio events in general, and defining anomalous/outlier sounds in particular. The application of such an anomaly detection problem to audio surveillance systems is made possible thanks to the advances of anomaly detection techniques, in particular for highly unbalanced data. However, outdoor audio signals are characterized by a high degree of uncertainty, since there is no way to model each category of sound, whether normal or anomalous, in presence of background noise. Thus, this paper proposes a rare/anomalous sound event detection method for road traffic surveillance, which aims at detecting hazardous events, such as car accidents, in presence of traffic noise. To model uncertainty for anomaly detection, the suggested method combines deep reconstruction techniques, interval-valued type-2 fuzzy sets and interval comparison methods. First, the reconstruction error of the input audio segment is yielded by a deep variational autoencoder which is trained on normal data only. Based on this reconstruction error, a fuzzy membership function with pessimistic/lower and optimistic/upper components is calculated. Next, a probabilistic interval comparison method is used to compute the membership score, and thus to evaluate the interval-valued fuzzy sets. Finally, defuzzification is used to classify events as normal or anomalous. During this process, several types of linear or nonlinear membership functions are utilized to model uncertainty with respect to the input, i.e., the VAE reconstruction error, or to the output, i.e., the value of the primary membership. According to the results obtained, the proposed method outperforms the state-of-the-art one-class SVM for anomaly detection and the baseline VAE error thresholding method, when specific parameters are carefully set, such as the weights of the anomalous/normal subsets and the lower/upper bounds of the membership function's components. Furthermore, the proposed linear and nonlinear membership functions succeed to improve modeling uncertainty in audio signals by interval-valued type-2 fuzzy sets, with regard to: a) the input, i.e., the VAE reconstruction error, and b) the primary membership value, respectively
A Novel Pitch Detection Algorithm Based on Instantaneous Frequency for Clean and Noisy Speech
No full textIn this paper, a novel pitch detection algorithm (PDA) is proposed. Actually, pitch detection is a classical problem that has been investigated since the very beginning of speech processing. However, the novelty of the proposed method consists in establishing an empirical relationship between fundamental frequency (f0) and instantaneous frequency (fi), which serves as a basis to develop the proposed PDA. Even though f0 and fi are defined as attributes of two different transforms, i.e., the Fourier transform and the Hilbert transform, respectively, the relationship proposed in this paper shows some interaction between both of them, at least empirically. The first step of this work consists in validating the proposed relationship on a large set of speech signals. Then, it is leveraged to develop an algorithm capable to (a) detect voiced/unvoiced parts of speech and (b) extract f0 contour from fi values in the voiced parts. For evaluation purposes, the yielding f0 contour is compared to some well-rated state-of-the-art PDA’s. The main findings show that the quality of pitch detection obtained by the proposed technique is as satisfactory as some of top PDA’s, either in clean or in simulated noisy speech. In addition, one of the main advantages consists in bypassing the traditional short-time analysis required to assume local stationarity in speech signal
Anomalous sound event detection: A survey of machine learning based methods and applications
No full text
Statistical Analysis of the Prosodic Parameters of a Spontaneous Arabic Speech Corpus for Speech Synthesis
No full text- …
