Abstract-We propose a flexible framework for spectral conversion (SC) that facilitates training with unaligned corpora. Many SC frameworks require parallel corpora, phonetic alignments, or explicit frame-wise correspondence for learning conversion functions or for synthesizing a target spectrum with the aid of alignments. However, these requirements gravely limit the scope of practical applications of SC due to scarcity or even unavailability of parallel corpora. We propose an SC framework based on variational auto-encoder which enables us to exploit non-parallel corpora. The framework comprises an encoder that learns speaker-independent phonetic representations and a decoder that learns to reconstruct the designated speaker. It removes the requirement of parallel corpora or phonetic alignments to train a spectral conversion system. We report objective and subjective evaluations to validate our proposed method and compare it to SC methods that have access to aligned corpora.

No abstract

Speech enhancement model is used to map a noisy speech to a clean speech. In the training stage, an objective function is often adopted to optimize the model parameters. However, in the existing literature, there is an inconsistency between the model optimization criterion and the evaluation criterion for the enhanced speech. For example, in measuring speech intelligibility, most of the evaluation metric is based on a short-time objective intelligibility (STOI) measure, while the frame based mean square error (MSE) between estimated and clean speech is widely used in optimizing the model. Due to the inconsistency, there is no guarantee that the trained model can provide optimal performance in applications. In this study, we propose an end-to-end utterance-based speech enhancement framework using fully convolutional neural networks (FCN) to reduce the gap between the model optimization and the evaluation criterion. Because of the utterance-based optimization, temporal correlation information of long speech segments, or even at the entire utterance level, can be considered to directly optimize perception-based objective functions. As an example, we implemented the proposed FCN enhancement framework to optimize the STOI measure. Experimental results show that the STOI of a test speech processed by the proposed approach is better than conventional MSE-optimized speech due to the consistency between the training and the evaluation targets. Moreover, by integrating the STOI into model optimization, the intelligibility of human subjects and automatic speech recognition system on the enhanced speech is also substantially improved compared to those generated based on the minimum MSE criterion.

No abstract

Building a voice conversion (VC) system from non-parallel speech corpora is challenging but highly valuable in real application scenarios. In most situations, the source and the target speakers do not repeat the same texts or they may even speak different languages. In this case, one possible, although indirect, solution is to build a generative model for speech. Generative models focus on explaining the observations with latent variables instead of learning a pairwise transformation function, thereby bypassing the requirement of speech frame alignment. In this paper, we propose a non-parallel VC framework with a variational autoencoding Wasserstein generative adversarial network (VAW-GAN) that explicitly considers a VC objective when building the speech model. Experimental results corroborate the capability of our framework for building a VC system from unaligned data, and demonstrate improved conversion quality. Index Terms: non-parallel voice conversion, Wasserstein generative adversarial network, GAN, variational autoencoder, VAE

No abstract

The electrocardiogram (ECG) is an efficient and noninvasive indicator for arrhythmia detection and prevention. In real-world scenarios, ECG signals are prone to be contaminated with various noises, which may lead to wrong interpretation. Therefore, significant attention has been paid on denoising of ECG for accurate diagnosis and analysis. A denoising autoencoder (DAE) can be applied to reconstruct the clean data from its noisy version. In this paper, a DAE using the fully convolutional network (FCN) is proposed for ECG signal denoising. Meanwhile, the proposed FCN-based DAE can perform compression with regard to the DAE architecture. The proposed approach is applied to ECG signals from the MIT-BIH Arrhythmia database and the added noise signals are obtained from the MIT-BIH Noise Stress Test database. The denoising performance is evaluated using the root-mean-square error (RMSE), percentage-root-mean-square difference (PRD), and improvement in signal-to-noise ratio (SNR imp ). The results of the experiments conducted on noisy ECG signals of different levels of input SNR show that the FCN acquires better performance as compared to the deep fully connected neural network-and convolutional neural network-based denoising models. Moreover, the proposed FCN-based DAE reduces the size of the input ECG signals, where the compressed data is 32 times smaller than the original. The results of the study demonstrate the superiority of FCN in denoising, with lower RMSE and PRD, as well as higher SNR imp . According to the results, we believe that the proposed FCN-based DAE has a good application prospect in clinical practice.INDEX TERMS Electrocardiography, signal denoising, artificial neural networks, denoising autoencoders, fully convolutional network.

This study proposes a fully convolutional network (FCN) model for raw waveform-based speech enhancement. The proposed system performs speech enhancement in an end-to-end (i.e., waveform-in and waveform-out) manner, which differs from most existing denoising methods that process the magnitude spectrum (e.g., log power spectrum (LPS)) only. Because the fully connected layers, which are involved in deep neural networks (DNN) and convolutional neural net-works (CNN), may not accurately characterize the local in-formation of speech signals, particularly with high frequency components, we employed fully convolutional layers to model the waveform. More specifically, FCN consists of only convolutional layers and thus the local temporal structures of speech signals can be efficiently and effectively preserved with relatively few weights. Experimental results show that DNN-and CNN-based models have limited capability to restore high frequency components of waveforms, thus leading to decreased intelligibility of enhanced speech. By contrast, the proposed FCN model can not only effectively recover the waveforms but also outperform the LPSbased DNN baseline in terms of short-time objective intelligibility (STOI) and perceptual evaluation of speech quality (PESQ). In addition, the number of model parameters in FCN is approximately only 0.2% com-pared with that in both DNN and CNN.I.