Papers by Keyword: Music

Abstract: Facial recognition based music system plays an important role in the treatment of human psychology. Face recognition system is an extensively used technique in most of the applications such as security system, video processing, in surveillance system and so on. People are often confused while choosing the kind of music they would want to listen. Relatively, this paper focuses on making an efficient music recommendation system which will recommend a suitable music to make the person feel sooth using Facial Recognition Techniques. This system uses FER-2013 dataset for training of the CNN, which is made using mini-xception architecture. Augmentation techniques are used for increasing the number of images in the dataset for training, which helps to increase the accuracy of the prediction. The face is captured using webcam and facial extraction is done using Haarcascade classifier and then sent to the CNN layers. The mini xception algorithm used in these CNN layers makes the system lighter and efficient as compared to existing systems. The accuracy of the proposed model is calculated and found to have reached the barrier threshold of 95% and average accuracy was found to be 90%. The song is recommended to the user using the proposed mapping algorithm.

Abstract: Based on estimation of signal parameters via singular value decomposition (SVD) filtering technique and Multiple Signal Classification (MUSIC) combined with extended Prony algorithm, a new method for induction motor’s broken rotor bar detection is proposed. Experiments show that the method is effective in applying to detect the fault of motor’s induction with the situation of interference noise and load fluctuations. And it is more practical in line detection because the algorithm is relatively simple and the running time is short.

Abstract: Brain computer interfaces (BCIs) have become a research hotspot in recent years because of great potentials to help disabled people communicate with the outside world. Among different paradigms, steady state visual evoked potential (SSVEP)-based BCIs are commonly implemented in real applications, because they provide higher signal to noise ratio (SNR) and greater information transfer rate (ITR) than other BCI techniques. Various algorithms have been employed for SSVEP signal processing, like fast Fourier transform (FFT), wavelet analysis and canonical correlation analysis (CCA). In this paper, a new method based on multiple signal classification (MUSIC) was proposed for SSVEP feature extraction. The experimental results proved that it could provide higher frequency resolution and the recognition accuracy was excellent via adjusting some parameters.

Abstract: In this thesis, fault characteristic frequency can not be accurately detected if they are drowned by noise. A novel method which is Multiple Signal Classification (MUSIC) based on four-order cumulate is provided to diagnose motor broken rotor bars and stator winding inter-turn short circuit. Because four-order accumulation is able to depress noise meanwhile fault information can be obtained accurately through this way presented even with small samples. Simulation results shown that the method were higher in resolution of frequency, more accurately in fault detection and less in computational complexity.

Abstract: To solve the problem of moving defect localization for wheel-bearings, a novel algorithm based on particle filter and multiple signal classification (MUSIC) is proposed in this paper. It introduces two-dimensional circular sensor array to measure acoustic signals of defective bearings. By through of MUSIC, the direction-of-arrivals (DOAs) of defective signal are firstly estimated. After the motion trajectory was calculated by particle filter and DOAs, the defect was located by reference sound source. The experimental results show that the radius and phase errors of proposed method are less than 2mm and 5 degrees.

Abstract: A platform of spatial estimation, which is based on Matlab GUI, was established. We use Matlab to do analysis of different parameters (Signal-to-noise ratio, the number of snapshots, the number of antenna elements, the number of targets) and simulate the results with different algorithm. In addition, we use a Matlab GUI to present the results of simulations in a interface-friendly way.

Abstract: The link between the origin and development of music, on one hand, and appearance and development of emotional intelligence, on the other hand, is considered. It is shown that at present the prerequisites exist to create a measuring instrument that could enable the level of emotions in acoustic signals to be recognized and evaluated. The prospects of future studies in the field are observed.

Abstract: This paper presents a method to detect weak harmonic signal embedded in chaotic noise. Using different correlation characteristic of harmonic and chaotic signal ,we can transform the sample signal to a new data sequence which has new harmonic . The new harmonic frequency is m times of the original harmonic and beyond the center bandwidth of noise. Then use wavelet packet decomposition to analysis the energy distribution of harmonic and chaotic signals and extract the component which the harmonic energy concentrated on, In the end, a multiple signal classification (MUSIC) algorithm is employed to estimate harmonic frequencies . The method suit for the complex background noise (strong chaotic noise and gaussian noise).

Abstract: Estimation for direction of arrival (DOA) is an important work in array signal processing, and subspace method such as MUSIC algorithm is basic and important in DOA estimation. This paper analyzes the structure of eigen value of variance matrix, and proposes a method to estimate the signal noise ratio (SNR) of the data received by sensor array. With the accurate estimation for SNR, we can estimate the work environment and decide detect threshold for many algorithm. The paper also proposes a method to promote the SNR of covariance matrix with moving the covariance slice to do DOA estimation. It can efficiently widen the difference of signal eigen value and noise eigen value.

Abstract: This paper focused on UM2000 signal spectrum estimation using MUSIC algorithm. Because of the limitation of data window length, traditional frequency discrimination methods fail to meet the requirement of high frequency resolution. In this paper, the influence of SNR on MUSIC spectrum estimation is analyzed and MDL (minimum description length) principle is used to determine the dimension of the signal. Simulation results based on several other modern spectral estimation methods are also presented and compared with that of MUSIC method, from which the superiority of MUSIC method is verified.