Automatic Speech Recognition (ASR) systems typically yield output in lexical form. However, humans prefer a written form output. To bridge this gap, ASR systems usually employ Inverse Text Normalization (ITN). In previous works, Weighted Finite State Transducers (WFST) have been employed to do ITN. WFSTs are nicely suited to this task but their size and run-time costs can make deployment on embedded applications challenging. In this paper, we describe the development of an on-device ITN system that is streaming, lightweight & accurate. At the core of our system is a streaming transformer tagger, that tags lexical tokens from ASR. The tag informs which ITN category might be applied, if at all. Following that, we apply an ITN-category-specific WFST, only on the tagged text, to reliably perform the ITN conversion. We show that the proposed ITN solution performs equivalent to strong baselines, while being significantly smaller in size and retaining customization capabilities.
Let $V_* : \mathbb{R}^d \to \mathbb{R}$ be some (possibly non-convex) potential function, and consider the probability measure $\pi \propto e^{-V_*}$. When $\pi$ exhibits multiple modes, it is known that sampling techniques based on Wasserstein gradient flows of the Kullback-Leibler (KL) divergence (e.g. Langevin Monte Carlo) suffer poorly in the rate of convergence, where the dynamics are unable to easily traverse between modes. In stark contrast, the work of Lu et al. (2019; 2022) has shown that the gradient flow of the KL with respect to the Fisher-Rao (FR) geometry exhibits a convergence rate to $\pi$ is that \textit{independent} of the potential function. In this short note, we complement these existing results in the literature by providing an explicit expansion of $\text{KL}(\rho_t^{\text{FR}}\|\pi)$ in terms of $e^{-t}$, where $(\rho_t^{\text{FR}})_{t\geq 0}$ is the FR gradient flow of the KL divergence. In turn, we are able to provide a clean asymptotic convergence rate, where the burn-in time is guaranteed to be finite. Our proof is based on observing a similarity between FR gradient flows and simulated annealing with linear scaling, and facts about cumulant generating functions. We conclude with simple synthetic experiments that demonstrate our theoretical findings are indeed tight. Based on our numerics, we conjecture that the asymptotic rates of convergence for Wasserstein-Fisher-Rao gradient flows are possibly related to this expansion in some cases.
Pretrained language models (PLMs) have shown marvelous improvements across various NLP tasks. Most Chinese PLMs simply treat an input text as a sequence of characters, and completely ignore word information. Although Whole Word Masking can alleviate this, the semantics in words is still not well represented. In this paper, we revisit the segmentation granularity of Chinese PLMs. We propose a mixed-granularity Chinese BERT (MigBERT) by considering both characters and words. To achieve this, we design objective functions for learning both character and word-level representations. We conduct extensive experiments on various Chinese NLP tasks to evaluate existing PLMs as well as the proposed MigBERT. Experimental results show that MigBERT achieves new SOTA performance on all these tasks. Further analysis demonstrates that words are semantically richer than characters. More interestingly, we show that MigBERT also works with Japanese. Our code has been released here~\footnote{\url{https://github.com/xnliang98/MigBERT}} and you can download our model here~\footnote{\url{https://huggingface.co/xnliang/MigBERT-large/}}.
Scene-text image synthesis techniques aimed at naturally composing text instances on background scene images are very appealing for training deep neural networks because they can provide accurate and comprehensive annotation information. Prior studies have explored generating synthetic text images on two-dimensional and three-dimensional surfaces based on rules derived from real-world observations. Some of these studies have proposed generating scene-text images from learning; however, owing to the absence of a suitable training dataset, unsupervised frameworks have been explored to learn from existing real-world data, which may not result in a robust performance. To ease this dilemma and facilitate research on learning-based scene text synthesis, we propose DecompST, a real-world dataset prepared using public benchmarks, with three types of annotations: quadrilateral-level BBoxes, stroke-level text masks, and text-erased images. Using the DecompST dataset, we propose an image synthesis engine that includes a text location proposal network (TLPNet) and a text appearance adaptation network (TAANet). TLPNet first predicts the suitable regions for text embedding. TAANet then adaptively changes the geometry and color of the text instance according to the context of the background. Our comprehensive experiments verified the effectiveness of the proposed method for generating pretraining data for scene text detectors.
We propose ClipFace, a novel self-supervised approach for text-guided editing of textured 3D morphable model of faces. Specifically, we employ user-friendly language prompts to enable control of the expressions as well as appearance of 3D faces. We leverage the geometric expressiveness of 3D morphable models, which inherently possess limited controllability and texture expressivity, and develop a self-supervised generative model to jointly synthesize expressive, textured, and articulated faces in 3D. We enable high-quality texture generation for 3D faces by adversarial self-supervised training, guided by differentiable rendering against collections of real RGB images. Controllable editing and manipulation are given by language prompts to adapt texture and expression of the 3D morphable model. To this end, we propose a neural network that predicts both texture and expression latent codes of the morphable model. Our model is trained in a self-supervised fashion by exploiting differentiable rendering and losses based on a pre-trained CLIP model. Once trained, our model jointly predicts face textures in UV-space, along with expression parameters to capture both geometry and texture changes in facial expressions in a single forward pass. We further show the applicability of our method to generate temporally changing textures for a given animation sequence.
The development of privacy-enhancing technologies has made immense progress in reducing trade-offs between privacy and performance in data exchange and analysis. Similar tools for structured transparency could be useful for AI governance by offering capabilities such as external scrutiny, auditing, and source verification. It is useful to view these different AI governance objectives as a system of information flows in order to avoid partial solutions and significant gaps in governance, as there may be significant overlap in the software stacks needed for the AI governance use cases mentioned in this text. When viewing the system as a whole, the importance of interoperability between these different AI governance solutions becomes clear. Therefore, it is imminently important to look at these problems in AI governance as a system, before these standards, auditing procedures, software, and norms settle into place.
Face aging is an ill-posed problem because multiple plausible aging patterns may correspond to a given input. Most existing methods often produce one deterministic estimation. This paper proposes a novel CLIP-driven Pluralistic Aging Diffusion Autoencoder (PADA) to enhance the diversity of aging patterns. First, we employ diffusion models to generate diverse low-level aging details via a sequential denoising reverse process. Second, we present Probabilistic Aging Embedding (PAE) to capture diverse high-level aging patterns, which represents age information as probabilistic distributions in the common CLIP latent space. A text-guided KL-divergence loss is designed to guide this learning. Our method can achieve pluralistic face aging conditioned on open-world aging texts and arbitrary unseen face images. Qualitative and quantitative experiments demonstrate that our method can generate more diverse and high-quality plausible aging results.
Zero-shot multi-speaker text-to-speech (ZSM-TTS) models aim to generate a speech sample with the voice characteristic of an unseen speaker. The main challenge of ZSM-TTS is to increase the overall speaker similarity for unseen speakers. One of the most successful speaker conditioning methods for flow-based multi-speaker text-to-speech (TTS) models is to utilize the functions which predict the scale and bias parameters of the affine coupling layers according to the given speaker embedding vector. In this letter, we improve on the previous speaker conditioning method by introducing a speaker-normalized affine coupling (SNAC) layer which allows for unseen speaker speech synthesis in a zero-shot manner leveraging a normalization-based conditioning technique. The newly designed coupling layer explicitly normalizes the input by the parameters predicted from a speaker embedding vector while training, enabling an inverse process of denormalizing for a new speaker embedding at inference. The proposed conditioning scheme yields the state-of-the-art performance in terms of the speech quality and speaker similarity in a ZSM-TTS setting.
While code-mixing is a common linguistic practice in many parts of the world, collecting high-quality and low-cost code-mixed data remains a challenge for natural language processing (NLP) research. The proliferation of Large Language Models (LLMs) in recent times compels one to ask: can these systems be used for data generation? In this article, we explore prompting multilingual LLMs in a zero-shot manner to create code-mixed data for five languages in South East Asia (SEA) -- Indonesian, Malay, Chinese, Tagalog, Vietnamese, as well as the creole language Singlish. We find that ChatGPT shows the most potential, capable of producing code-mixed text 68% of the time when the term "code-mixing" is explicitly defined. Moreover, both ChatGPT's and InstructGPT's (davinci-003) performances in generating Singlish texts are noteworthy, averaging a 96% success rate across a variety of prompts. Their code-mixing proficiency, however, is dampened by word choice errors that lead to semantic inaccuracies. Other multilingual models such as BLOOMZ and Flan-T5-XXL are unable to produce code-mixed texts altogether. By highlighting the limited promises of LLMs in a specific form of low-resource data generation, we call for a measured approach when applying similar techniques to other data-scarce NLP contexts.
Image editing using diffusion models has witnessed extremely fast-paced growth recently. There are various ways in which previous works enable controlling and editing images. Some works use high-level conditioning such as text, while others use low-level conditioning. Nevertheless, most of them lack fine-grained control over the properties of the different objects present in the image, i.e. object-level image editing. In this work, we consider an image as a composition of multiple objects, each defined by various properties. Out of these properties, we identify structure and appearance as the most intuitive to understand and useful for editing purposes. We propose Structure-and-Appearance Paired Diffusion model (PAIR-Diffusion), which is trained using structure and appearance information explicitly extracted from the images. The proposed model enables users to inject a reference image's appearance into the input image at both the object and global levels. Additionally, PAIR-Diffusion allows editing the structure while maintaining the style of individual components of the image unchanged. We extensively evaluate our method on LSUN datasets and the CelebA-HQ face dataset, and we demonstrate fine-grained control over both structure and appearance at the object level. We also applied the method to Stable Diffusion to edit any real image at the object level.