Shijun Wang

Shijun Wang, Jón Guðnason, Damian Borth · berkeley

State-of-the-art Text-To-Speech (TTS) models are capable of producing high-quality speech. The generated speech, however, is usually neutral in emotional expression, whereas very often one would want fine-grained emotional control of words or phonemes. Although still challenging, the first TTS models have been recently proposed that are able to control voice by manually assigning emotion intensity. Unfortunately, due to the neglect of intra-class distance, the intensity differences are often unrecognizable. In this paper, we propose a fine-grained controllable emotional TTS, that considers both inter- and intra-class distances and be able to synthesize speech with recognizable intensity difference. Our subjective and objective experiments demonstrate that our model exceeds two state-of-the-art controllable TTS models for controllability, emotion expressiveness and naturalness.

Shijun Wang, Jón Guðnason, Damian Borth · berkeley

Effective speech emotional representations play a key role in Speech Emotion Recognition (SER) and Emotional Text-To-Speech (TTS) tasks. However, emotional speech samples are more difficult and expensive to acquire compared with Neutral style speech, which causes one issue that most related works unfortunately neglect: imbalanced datasets. Models might overfit to the majority Neutral class and fail to produce robust and effective emotional representations. In this paper, we propose an Emotion Extractor to address this issue. We use augmentation approaches to train the model and enable it to extract effective and generalizable emotional representations from imbalanced datasets. Our empirical results show that (1) for the SER task, the proposed Emotion Extractor surpasses the state-of-the-art baseline on three imbalanced datasets; (2) the produced representations from our Emotion Extractor benefit the TTS model, and enable it to synthesize more expressive speech.

Siqiao Xue, Chao Qu, Xiaoming Shi et al.

Predictive autoscaling (autoscaling with workload forecasting) is an important mechanism that supports autonomous adjustment of computing resources in accordance with fluctuating workload demands in the Cloud. In recent works, Reinforcement Learning (RL) has been introduced as a promising approach to learn the resource management policies to guide the scaling actions under the dynamic and uncertain cloud environment. However, RL methods face the following challenges in steering predictive autoscaling, such as lack of accuracy in decision-making, inefficient sampling and significant variability in workload patterns that may cause policies to fail at test time. To this end, we propose an end-to-end predictive meta model-based RL algorithm, aiming to optimally allocate resource to maintain a stable CPU utilization level, which incorporates a specially-designed deep periodic workload prediction model as the input and embeds the Neural Process to guide the learning of the optimal scaling actions over numerous application services in the Cloud. Our algorithm not only ensures the predictability and accuracy of the scaling strategy, but also enables the scaling decisions to adapt to the changing workloads with high sample efficiency. Our method has achieved significant performance improvement compared to the existing algorithms and has been deployed online at Alipay, supporting the autoscaling of applications for the world-leading payment platform.

Zhonghua Liu, Shijun Wang, Ning Chen

Voice Conversion (VC) converts the voice of a source speech to that of a target while maintaining the source's content. Speech can be mainly decomposed into four components: content, timbre, rhythm and pitch. Unfortunately, most related works only take into account content and timbre, which results in less natural speech. Some recent works are able to disentangle speech into several components, but they require laborious bottleneck tuning or various hand-crafted features, each assumed to contain disentangled speech information. In this paper, we propose a VC model that can automatically disentangle speech into four components using only two augmentation functions, without the requirement of multiple hand-crafted features or laborious bottleneck tuning. The proposed model is straightforward yet efficient, and the empirical results demonstrate that our model can achieve a better performance than the baseline, regarding disentanglement effectiveness and speech naturalness.

Shijun Wang, Dimche Kostadinov, Damian Borth

Voice Conversion (VC) for unseen speakers, also known as zero-shot VC, is an attractive research topic as it enables a range of applications like voice customizing, animation production, and others. Recent work in this area made progress with disentanglement methods that separate utterance content and speaker characteristics from speech audio recordings. However, many of these methods are subject to the leakage of prosody (e.g., pitch, volume), causing the speaker voice in the synthesized speech to be different from the desired target speakers. To prevent this issue, we propose a novel self-supervised approach that effectively learns disentangled pitch and volume representations that can represent the prosody styles of different speakers. We then use the learned prosodic representations as conditional information to train and enhance our VC model for zero-shot conversion. In our experiments, we show that our prosody representations are disentangled and rich in prosody information. Moreover, we demonstrate that the addition of our prosody representations improves our VC performance and surpasses state-of-the-art zero-shot VC performances.

Shijun Wang, Damian Borth

Voice conversion (VC) is a task that transforms voice from target audio to source without losing linguistic contents, it is challenging especially when source and target speakers are unseen during training (zero-shot VC). Previous approaches require a pre-trained model or linguistic data to do the zero-shot conversion. Meanwhile, VC models with Vector Quantization (VQ) or Instance Normalization (IN) are able to disentangle contents from audios and achieve successful conversions. However, disentanglement in these models highly relies on heavily constrained bottleneck layers, thus, the sound quality is drastically sacrificed. In this paper, we propose NoiseVC, an approach that can disentangle contents based on VQ and Contrastive Predictive Coding (CPC). Additionally, Noise Augmentation is performed to further enhance disentanglement capability. We conduct several experiments and demonstrate that NoiseVC has a strong disentanglement ability with a small sacrifice of quality.

Baocheng Zhu, Shijun Wang, James Zhang

We present a novel architecture named Neural Physicist (NeurPhy) to learn physical dynamics directly from image sequences using deep neural networks. For any physical system, given the global system parameters, the time evolution of states is governed by the underlying physical laws. How to learn meaningful system representations in an end-to-end way and estimate accurate state transition dynamics facilitating long-term prediction have been long-standing challenges. In this paper, by leveraging recent progresses in representation learning and state space models (SSMs), we propose NeurPhy, which uses variational auto-encoder (VAE) to extract underlying Markovian dynamic state at each time step, neural process (NP) to extract the global system parameters, and a non-linear non-recurrent stochastic state space model to learn the physical dynamic transition. We apply NeurPhy to two physical experimental environments, i.e., damped pendulum and planetary orbits motion, and achieve promising results. Our model can not only extract the physically meaningful state representations, but also learn the state transition dynamics enabling long-term predictions for unseen image sequences. Furthermore, from the manifold dimension of the latent state space, we can easily identify the degree of freedom (DoF) of the underlying physical systems.

Shijun Wang, Baocheng Zhu, Chen Li et al.

In this paper, We propose a general Riemannian proximal optimization algorithm with guaranteed convergence to solve Markov decision process (MDP) problems. To model policy functions in MDP, we employ Gaussian mixture model (GMM) and formulate it as a nonconvex optimization problem in the Riemannian space of positive semidefinite matrices. For two given policy functions, we also provide its lower bound on policy improvement by using bounds derived from the Wasserstein distance of GMMs. Preliminary experiments show the efficacy of our proposed Riemannian proximal policy optimization algorithm.

Shijun Wang, Baocheng Zhu, Lintao Ma et al.

In this paper, we consider optimizing a smooth, convex, lower semicontinuous function in Riemannian space with constraints. To solve the problem, we first convert it to a dual problem and then propose a general primal-dual algorithm to optimize the primal and dual variables iteratively. In each optimization iteration, we employ a proximal operator to search optimal solution in the primal space. We prove convergence of the proposed algorithm and show its non-asymptotic convergence rate. By utilizing the proposed primal-dual optimization technique, we propose a novel metric learning algorithm which learns an optimal feature transformation matrix in the Riemannian space of positive definite matrices. Preliminary experimental results on an optimal fund selection problem in fund of funds (FOF) management for quantitative investment showed its efficacy.