Search

In this paper, the problems of finite-time stability and stabilization for switched positive linear time-delay systems under mode-dependent average dwell time (MDADT) are investigated. By proposing a novel multiple piecewise copositive Lyapunov-Krasovskii functional, new results on the sufficient conditions of finite-time stability are obtained. In order to solve the limitation of the multiple piecewise copositive Lyapunov-Krasovskii functional on designing controllers, a novel linear combinatorial copositive Lyapunov-Krasovskii functional is constructed, this provides a possibility for the numerical construction of the controller. Then by using state feedback controller, the finite-time stabilization is achieved. Finally, some simulation results are given to show the advantages of our methods.

This paper presents methods of making using of text supervision to improve the performance of sequence-to-sequence (seq2seq) voice conversion. Compared with conventional frame-to-frame voice conversion approaches, the seq2seq acoustic modeling method proposed in our previous work achieved higher naturalness and similarity. In this paper, we further improve its performance by utilizing the text transcriptions of parallel training data. First, a multi-task learning structure is designed which adds auxiliary classifiers to the middle layers of the seq2seq model and predicts linguistic labels as a secondary task. Second, a data-augmentation method is proposed which utilizes text alignment to produce extra parallel sequences for model training. Experiments are conducted to evaluate our proposed method with training sets at different sizes. Experimental results show that the multi-task learning with linguistic labels is effective at reducing the errors of seq2seq voice conversion. The data-augmentation method can further improve the performance of seq2seq voice conversion when only 50 or 100 training utterances are available.

The Blizzard Machine Learning Challenge (BMLC) aims to liberate participants from speech-specific processing when building speech synthesis systems. This paper describes the USTC system for the ES2 sub-task in BMLC2017, which requires participants to train a model to directly predict waveforms from linguistic features. We investigate three aspects of waveform modeling when preparing our system for this task. First, two different model structures for waveform modeling, i.e., WaveNet and SampleRNN, are compared on this task. Second, a strategy of using features extracted from waveforms as intermediate representations for waveform modeling is studied. Experimental results show that using low-level features (STFT amplitude spectra) as intermediate representations can achieve similar performance as using high-level features (mel-cepstra and F0). Third, the feasibility of applying WaveNet to wideband speech signals with more than 256 quantization levels is verified by experiments. Finally, a system which adopts STFT amplitude spectra as intermediate representations to model 24kHz speech waveforms with 1024 mu-law quantization levels is submitted for evaluation. The evaluation results of BMLC2017 demonstrate the effectiveness of our proposed methods.

This paper introduces the speech synthesis system submitted by IFLYTEK for the Blizzard Machine Learning Challenge 2017-ES1. Linguistic and acoustic features from a 4hour corpus were released for this task. Participants are expected to build a speech synthesis system on the given linguist and acoustic features without using any external data. Our system is composed of a long short term memory (LSTM) recurrent neural network (RNN)-based acoustic model and a generative adversarial network (GAN)-based post-filter for mel-cepstra. Two approaches to build GAN-based post-filter are implemented and compared in our experiments. The first one is to predict the residuals of mel-cepstra given the mel-cepstra predicted by the LSTM-based acoustic model. However, this method leads to unstable synthetic speech sounds in our experiments, which may be due to the poor quality of analysis-synthesis speech using the natural acoustic features given by this corpus. The other approach is to ignore the detailed components of natural mel-cepstra by dimension reduction using principal component analysis (PCA) and then recover them back using GAN given the main PCA components. At synthesis time, mel-cepstra predicted by the RNN acoustic model are first projected to the main PCA components, which are then sent to the GAN for detail recovering. Finally, the second approach is used in the final submitted system. The evaluation results show the effectiveness of our submitted system.

In this paper, the problem of positive observer design of linear positive systems is studied. This paper aims to providing new design approaches for positive observers of linear positive systems. First, an efficient continuous-time positive observer design approach is proposed by using linear programming such that the observer error system is exponentially stable. Then, the results are also extended to discrete-time positive time-varying delay systems. Finally, two comparison examples are given to show the merit of the new design approach.

In this paper, the problem of adaptive backstepping control for uncertain systems in the presence of unmodeled dynamics and input time-varying delays is studied. Under some mild assumptions, a robust adaptive controller is designed such that the system is globally stabilized by using adaptive backstepping technique. Meanwhile, the transient system performance in L2 and norms of system output can be adjusted by choosing the design parameters. Finally, a simulation example is given to show the effectiveness of the results.

The spectral envelope is the most natural representation of speech signal. But in voice conversion, it is difficult to directly model the raw spectral envelope space, which is high dimensional and strongly cross-dimensional correlated, with conventional Gaussian distributions. Bidirectional associative memory (BAM) is a two-layer feedback neural network that can better model the cross-dimensional correlations in high dimensional vectors. In this paper, we propose to reformulate BAMs as Gaussian distributions in order to model the spectral envelope space. The parameters of BAMs are estimated using the contrastive divergence algorithm. The evaluations on likelihood show that BAMs have better modeling ability than Gaussians with diagonal covariance. And the subjective tests on voice conversion indicate that the performance of the proposed method is significantly improved comparing with the conventional GMM based method.

When fatal forest disasters take place, only one retrieval depot usually can not provide enough supplies to resolve the emergency. So the problem about how to combine multi-depot-multi-resource has been put forward. The main objective of the multi-resource scheduling is how to achieve the least emergency time and the fewest retrieval depots. In this scheduling algorithm, there are mainly two factors that restrict the emergency time. One is path length, and the other is traffic road quality. Because the forest disaster usually takes place in remote mountain area, the factor of traffic road quality is very important. Therefore the least time (tij) should be the optimal solution about the shortest path (dij) and the road quality parameter (qij). In this paper, taking a forest area as experiments region, we achieved some results mainly in the following aspects: (1) determined the value of qij through the experiment; (2) established the least emergency time model; (3) built the emergency resources scheduling optimal model in forest fatal disaster for meeting fewest rescue depots based on the fastest emergency time.

This paper proposes an image recognition method, which consists of two steps: features extraction based on wavelet transform and image recognition using artificial neural networks. More specifically, wavelet transform is used to decompose the original image into different frequency sub-bands, then a set of features are extracted from the wavelet coefficients. The feature set as input fed into neural network for recognition. The experimental results confirmed effectiveness of the proposed approach.