We propose a new method for music detection from broadcasting contents using the convolutional neural networks with a Mel-scale kernel. In this detection task, music segments should be annotated from the broadcast data, where music, speech, and noise are mixed. The convolutional neural network is composed of a convolutional layer with kernel that is trained to extract robust features. The Mel-scale changes the kernel size, and the backpropagation algorithm trains the kernel shape. We used 52 h of mixed broadcast data (25 h of music) to train the convolutional network and 24 h of collected broadcast data (ratio of music of 50-76%) for testing. The test data consisted of various genres (drama, documentary, news, kids, reality, and so on) that are broadcast in British English, Spanish, and Korean languages. The proposed method consistently showed better performance in all the three languages than the baseline system, and the F-score ranged from 86.5% for British data to 95.9% for Korean drama data. Our music detection system takes about 28 s to process a 1-min signal using only one CPU with 4 cores.

In this paper, we propose an automatic fluency evaluation algorithm for English speaking tests. In the proposed algorithm, acoustic features are extracted from an input spoken utterance and then fluency score is computed by using support vector regression (SVR). We estimate the parameters of feature modeling and SVR using the speech signals and the corresponding scores by human raters. From the correlation analysis results, it is shown that speech rate, articulation rate, and mean length of runs are best for fluency evaluation. Experimental results show that the correlation between the human score and the SVR score is 0.87 for 3 speaking tests, which suggests the possibility of the proposed algorithm as a secondary fluency evaluation tool. 1) 적은 묵음(pause)으로 구체적으로 말할 수 있는 능력 2) 응집력 있고 논리적으로 말하는 능력3) 다양한 문맥과 상황에서 적절히 말할 수 있는 능력

In this paper, we propose a deep learning architecture that can effectively detect speech segmentation in broadcast contents. We also propose a multi-scale time-dilated layer for learning the temporal changes of feature vectors. We implement several comparison models to verify the performance of proposed model and calculated the frame-by-frame F-score, precision, and recall. Both the proposed model and the comparison model are trained with the same training data, and we train the model using 32 hours of Korean broadcast data which is composed of various genres (drama, news, documentary, and so on). Our proposed model shows the best performance with F-score 91.7% in Korean broadcast data. The British and Spanish broadcast data also show the highest performance with F-score 87.9% and 92.6%. As a result, our proposed model can contribute to the improvement of performance of speech detection by learning the temporal changes of the feature vectors.

In the previous works on fluency, Fillmore defined the 4 elements of fluency: the ability to talk at length with minimal pauses, the ability to talk cohesively and logically, the ability to talk in a wide range of contexts or situations, and the ability to create talk [4]. Crystal defined the fluency as 'smooth, rapid, effortless use of language' [5]. Chamber established the definition of fluency in qualitative and quantitative aspects and proposed the evaluation guide for foreign language speaking tests. Chamber's experiments showed that the important elements for fluency evaluation are the rate of speech, the frequency or position of pause, and hesitations, which are temporal and quantitative features [2]. Kormos investigated the effects of temporal and lexical features on fluency evaluation and asserted that important features are the speech rate, the phonation time ratio, the number of stressed words, and the accuracy [3]. In the Deshmukh et al.'s study, 8 prosodic and 8 lexical features were extracted for fluency evaluation, and good performance was generally achieved with the lexical features among which the

We propose to use independent component analysis (ICA) and deep neural network (DNN) to detect music sections in broadcast drama contents. Drama contents mainly comprise silence, noise, speech, music, and mixed (speech+music) sections. The silence section is detected by signal activity detection. To detect the music section, we train noise, speech, music, and mixed models with DNN. In computer experiments, we used the MUSAN corpus for training the acoustic model, and conducted an experiment using 3 hours' worth of Korean drama contents. As the mixed section includes music signals, it was regarded as a music section. The segmentation error rate (SER) of music section detection was observed to be 19.0%. In addition, when stereo mixed signals were separated into music signals using ICA, the SER was reduced to 11.8%.