Abstract Automatic speaker recognition (ASR) is based on speech feature vectors, models, and classifiers. To improve the speaker recognition performance, we must affect at least one of these modules. In this paper we propose to use subband spectral centroids (SSCs) as a complementary features with the traditional MFCC features, and a new GMM training algorithm, with the ultimate goal to search... Automatic speaker recognition (ASR) is based on speech feature vectors, models, and classifiers. To improve the speaker recognition performance, we must affect at least one of these modules. In this paper we propose to use subband spectral centroids (SSCs) as a complementary features with the traditional MFCC features, and a new GMM training algorithm, with the ultimate goal to search the better mixture component number N for each speaker model, which is fixed in the most speaker recognition systems based on GMM without any priori information, and all speaker models have the same number of components. In experiments, we compared the performance of the proposed scheme with the conventional GMM to show its robustness. Read More Read Less

Abstract We develop in this work a speaker verification system depending on Arabic syllables by the study of the visual speech that contains the visual and acoustic modalities. The visual signal provides both additional information that is not present in the audio and also a visual representation of some of the information that is present in the audio. This is particularly... We develop in this work a speaker verification system depending on Arabic syllables by the study of the visual speech that contains the visual and acoustic modalities. The visual signal provides both additional information that is not present in the audio and also a visual representation of some of the information that is present in the audio. This is particularly evident in the perception of speech where the articulatory gestures of the speaker’s lips and face can significantly improve the listener’s detection. In order to analyze Arabic visual speech, we extract features such as pitch and LPC coefficients for the acoustic modality, and we use DCT coefficients for lip images for visual modality. Hierarchical ascendant classification (HAC) is applied for each modality. Simulation results show good recognition rate of speaker verification depending on phoneme for the two modalities. Read More Read Less

Abstract Warehousing projects can know cases of failure because they did not take into consideration the users’ needs especially if they are not experienced with the technologies of data warehouses. The solution consists on building the data warehouse incrementally by designing and implementing one data mart at a time. In this work we propose a method to design data mart schema... Warehousing projects can know cases of failure because they did not take into consideration the users’ needs especially if they are not experienced with the technologies of data warehouses. The solution consists on building the data warehouse incrementally by designing and implementing one data mart at a time. In this work we propose a method to design data mart schema from OLAP requirements that are presented as schemas and grouped according to their domain. We apply the data integration technique to merge the different schemas so that we get one data mart schema by group. The integration is composed by schema matching (detecting semantic correspondence and the conflicts) and schema mapping (solving the existing conflict). Read More Read Less

Abstract In this paper, a new hybrid algorithm which combines both of token-based and character-based approaches is presented. The basic Levenshtein approach has been extended to token-based distance metric. The distance metric is enhanced to set the proper granularity level behavior of the algorithm. It smoothly maps a threshold of misspellings differences at the character level, and the importance of token... In this paper, a new hybrid algorithm which combines both of token-based and character-based approaches is presented. The basic Levenshtein approach has been extended to token-based distance metric. The distance metric is enhanced to set the proper granularity level behavior of the algorithm. It smoothly maps a threshold of misspellings differences at the character level, and the importance of token level errors in terms of token’s position and frequency. Using a large Arabic dataset, the experimental results show that the proposed algorithm overcomes successfully many types of errors such as: typographical errors, omission or insertion of middle name components, omission of non-significant popular name components, and different writing styles character variations. When compared the results with other classical algorithms, using the same dataset, the proposed algorithm was found to increase the minimum success level of best tested algorithms, while achieving higher upperlimits. Read More Read Less