In this paper, we propose a differentiable WORLD synthesizer and demonstrate its use in end-to-end audio style transfer tasks such as (singing) voice conversion and the DDSP timbre transfer task. Accordingly, our baseline differentiable synthesizer has no model parameters, yet it yields adequate synthesis quality. We can extend the baseline synthesizer by appending lightweight black-box postnets which apply further processing to the baseline output in order to improve fidelity. An alternative differentiable approach considers extraction of the source excitation spectrum directly, which can improve naturalness albeit for a narrower class of style transfer applications. The acoustic feature parameterization used by our approaches has the added benefit that it naturally disentangles pitch and timbral information so that they can be modeled separately. Moreover, as there exists a robust means of estimating these acoustic features from monophonic audio sources, it allows for parameter loss terms to be added to an end-to-end objective function, which can help convergence and/or further stabilize (adversarial) training.
相關內容
Recent years have seen many successful applications of machine learning (ML) to facilitate fluid dynamic computations. As simulations grow, generating new training datasets for traditional offline learning creates I/O and storage bottlenecks. Additionally, performing inference at runtime requires non-trivial coupling of ML framework libraries with simulation codes. This work offers a solution to both limitations by simplifying this coupling and enabling in situ training and inference workflows on heterogeneous clusters. Leveraging SmartSim, the presented framework deploys a database to store data and ML models in memory, thus circumventing the file system. On the Polaris supercomputer, we demonstrate perfect scaling efficiency to the full machine size of the data transfer and inference costs thanks to a novel co-located deployment of the database. Moreover, we train an autoencoder in situ from a turbulent flow simulation, showing that the framework overhead is negligible relative to a solver time step and training epoch.
Modeling the spatial radiance distribution of light rays in a scene has been extensively explored for applications, including view synthesis. Spectrum and polarization, the wave properties of light, are often neglected due to their integration into three RGB spectral bands and their non-perceptibility to human vision. Despite this, these properties encompass substantial material and geometric information about a scene. In this work, we propose to model spectro-polarimetric fields, the spatial Stokes-vector distribution of any light ray at an arbitrary wavelength. We present Neural Spectro-polarimetric Fields (NeSpoF), a neural representation that models the physically-valid Stokes vector at given continuous variables of position, direction, and wavelength. NeSpoF manages inherently noisy raw measurements, showcases memory efficiency, and preserves physically vital signals, factors that are crucial for representing the high-dimensional signal of a spectro-polarimetric field. To validate NeSpoF, we introduce the first multi-view hyperspectral-polarimetric image dataset, comprised of both synthetic and real-world scenes. These were captured using our compact hyperspectral-polarimetric imaging system, which has been calibrated for robustness against system imperfections. We demonstrate the capabilities of NeSpoF on diverse scenes.
Variational regularization is commonly used to solve linear inverse problems, and involves augmenting a data fidelity by a regularizer. The regularizer is used to promote a priori information, and is weighted by a regularization parameter. Selection of an appropriate regularization parameter is critical, with various choices leading to very different reconstructions. Existing strategies such as the discrepancy principle and L-curve can be used to determine a suitable parameter value, but in recent years a supervised machine learning approach called bilevel learning has been employed. Bilevel learning is a powerful framework to determine optimal parameters, and involves solving a nested optimisation problem. While previous strategies enjoy various theoretical results, the well-posedness of bilevel learning in this setting is still a developing field. One necessary property is positivity of the determined regularization parameter. In this work, we provide a new condition that better characterises positivity of optimal regularization parameters than the existing theory. Numerical results verify and explore this new condition for both small and large dimensional problems.
Audio-visual speech recognition (AVSR) provides a promising solution to ameliorate the noise-robustness of audio-only speech recognition with visual information. However, most existing efforts still focus on audio modality to improve robustness considering its dominance in AVSR task, with noise adaptation techniques such as front-end denoise processing. Though effective, these methods are usually faced with two practical challenges: 1) lack of sufficient labeled noisy audio-visual training data in some real-world scenarios and 2) less optimal model generality to unseen testing noises. In this work, we investigate the noise-invariant visual modality to strengthen robustness of AVSR, which can adapt to any testing noises while without dependence on noisy training data, a.k.a., unsupervised noise adaptation. Inspired by human perception mechanism, we propose a universal viseme-phoneme mapping (UniVPM) approach to implement modality transfer, which can restore clean audio from visual signals to enable speech recognition under any noisy conditions. Extensive experiments on public benchmarks LRS3 and LRS2 show that our approach achieves the state-of-the-art under various noisy as well as clean conditions. In addition, we also outperform previous state-of-the-arts on visual speech recognition task.
The utilization of discrete speech tokens, divided into semantic tokens and acoustic tokens, has been proven superior to traditional acoustic feature mel-spectrograms in terms of naturalness and robustness for text-to-speech (TTS) synthesis. Recent popular models, such as VALL-E and SPEAR-TTS, allow zero-shot speaker adaptation through auto-regressive (AR) continuation of acoustic tokens extracted from a short speech prompt. However, these AR models are restricted to generate speech only in a left-to-right direction, making them unsuitable for speech editing where both preceding and following contexts are provided. Furthermore, these models rely on acoustic tokens, which have audio quality limitations imposed by the performance of audio codec models. In this study, we propose a unified context-aware TTS framework called UniCATS, which is capable of both speech continuation and editing. UniCATS comprises two components, an acoustic model CTX-txt2vec and a vocoder CTX-vec2wav. CTX-txt2vec employs contextual VQ-diffusion to predict semantic tokens from the input text, enabling it to incorporate the semantic context and maintain seamless concatenation with the surrounding context. Following that, CTX-vec2wav utilizes contextual vocoding to convert these semantic tokens into waveforms, taking into consideration the acoustic context. Our experimental results demonstrate that CTX-vec2wav outperforms HifiGAN and AudioLM in terms of speech resynthesis from semantic tokens. Moreover, we show that UniCATS achieves state-of-the-art performance in both speech continuation and editing.
Implicit Neural Representations (INRs) are nowadays used to represent multimedia signals across various real-life applications, including image super-resolution, image compression, or 3D rendering. Existing methods that leverage INRs are predominantly focused on visual data, as their application to other modalities, such as audio, is nontrivial due to the inductive biases present in architectural attributes of image-based INR models. To address this limitation, we introduce HyperSound, the first meta-learning approach to produce INRs for audio samples that leverages hypernetworks to generalize beyond samples observed in training. Our approach reconstructs audio samples with quality comparable to other state-of-the-art models and provides a viable alternative to contemporary sound representations used in deep neural networks for audio processing, such as spectrograms.
Classic algorithms and machine learning systems like neural networks are both abundant in everyday life. While classic computer science algorithms are suitable for precise execution of exactly defined tasks such as finding the shortest path in a large graph, neural networks allow learning from data to predict the most likely answer in more complex tasks such as image classification, which cannot be reduced to an exact algorithm. To get the best of both worlds, this thesis explores combining both concepts leading to more robust, better performing, more interpretable, more computationally efficient, and more data efficient architectures. The thesis formalizes the idea of algorithmic supervision, which allows a neural network to learn from or in conjunction with an algorithm. When integrating an algorithm into a neural architecture, it is important that the algorithm is differentiable such that the architecture can be trained end-to-end and gradients can be propagated back through the algorithm in a meaningful way. To make algorithms differentiable, this thesis proposes a general method for continuously relaxing algorithms by perturbing variables and approximating the expectation value in closed form, i.e., without sampling. In addition, this thesis proposes differentiable algorithms, such as differentiable sorting networks, differentiable renderers, and differentiable logic gate networks. Finally, this thesis presents alternative training strategies for learning with algorithms.
The conjoining of dynamical systems and deep learning has become a topic of great interest. In particular, neural differential equations (NDEs) demonstrate that neural networks and differential equation are two sides of the same coin. Traditional parameterised differential equations are a special case. Many popular neural network architectures, such as residual networks and recurrent networks, are discretisations. NDEs are suitable for tackling generative problems, dynamical systems, and time series (particularly in physics, finance, ...) and are thus of interest to both modern machine learning and traditional mathematical modelling. NDEs offer high-capacity function approximation, strong priors on model space, the ability to handle irregular data, memory efficiency, and a wealth of available theory on both sides. This doctoral thesis provides an in-depth survey of the field. Topics include: neural ordinary differential equations (e.g. for hybrid neural/mechanistic modelling of physical systems); neural controlled differential equations (e.g. for learning functions of irregular time series); and neural stochastic differential equations (e.g. to produce generative models capable of representing complex stochastic dynamics, or sampling from complex high-dimensional distributions). Further topics include: numerical methods for NDEs (e.g. reversible differential equations solvers, backpropagation through differential equations, Brownian reconstruction); symbolic regression for dynamical systems (e.g. via regularised evolution); and deep implicit models (e.g. deep equilibrium models, differentiable optimisation). We anticipate this thesis will be of interest to anyone interested in the marriage of deep learning with dynamical systems, and hope it will provide a useful reference for the current state of the art.
Geometry and shape are fundamental aspects of visual style. Existing style transfer methods focus on texture-like components of style, ignoring geometry. We propose deformable style transfer (DST), an optimization-based approach that integrates texture and geometry style transfer. Our method is the first to allow geometry-aware stylization not restricted to any domain and not requiring training sets of matching style/content pairs. We demonstrate our method on a diverse set of content and style images including portraits, animals, objects, scenes, and paintings.
We introduce an effective model to overcome the problem of mode collapse when training Generative Adversarial Networks (GAN). Firstly, we propose a new generator objective that finds it better to tackle mode collapse. And, we apply an independent Autoencoders (AE) to constrain the generator and consider its reconstructed samples as "real" samples to slow down the convergence of discriminator that enables to reduce the gradient vanishing problem and stabilize the model. Secondly, from mappings between latent and data spaces provided by AE, we further regularize AE by the relative distance between the latent and data samples to explicitly prevent the generator falling into mode collapse setting. This idea comes when we find a new way to visualize the mode collapse on MNIST dataset. To the best of our knowledge, our method is the first to propose and apply successfully the relative distance of latent and data samples for stabilizing GAN. Thirdly, our proposed model, namely Generative Adversarial Autoencoder Networks (GAAN), is stable and has suffered from neither gradient vanishing nor mode collapse issues, as empirically demonstrated on synthetic, MNIST, MNIST-1K, CelebA and CIFAR-10 datasets. Experimental results show that our method can approximate well multi-modal distribution and achieve better results than state-of-the-art methods on these benchmark datasets. Our model implementation is published here: //github.com/tntrung/gaan
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191