Xiang Hao

Xiang Hao, Jibin Wu, Jianwei Yu et al.

Humans can easily isolate a single speaker from a complex acoustic environment, a capability referred to as the "Cocktail Party Effect." However, replicating this ability has been a significant challenge in the field of target speaker extraction (TSE). Traditional TSE approaches predominantly rely on voiceprints, which raise privacy concerns and face issues related to the quality and availability of enrollment samples, as well as intra-speaker variability. To address these issues, this work introduces a novel text-guided TSE paradigm named LLM-TSE. In this paradigm, a state-of-the-art large language model, LLaMA 2, processes typed text input from users to extract semantic cues. We demonstrate that textual descriptions alone can effectively serve as cues for extraction, thus addressing privacy concerns and reducing dependency on voiceprints. Furthermore, our approach offers flexibility by allowing the user to specify the extraction or suppression of a speaker and enhances robustness against intra-speaker variability by incorporating context-dependent textual information. Experimental results show competitive performance with text-based cues alone and demonstrate the effectiveness of using text as a task selector. Additionally, they achieve a new state-of-the-art when combining text-based cues with pre-registered cues. This work represents the first integration of LLMs with TSE, potentially establishing a new benchmark in solving the cocktail party problem and expanding the scope of TSE applications by providing a versatile, privacy-conscious solution.

Zheyu Zhang, Ziqi Pang, Shixing Chen et al.

Long video understanding is inherently challenging for vision-language models (VLMs) because of the extensive number of frames. With each video frame typically expanding into tens or hundreds of tokens, the limited context length of large language models (LLMs) forces the VLMs to perceive the frames sparsely and lose temporal information. To address this, we explore extreme video token compression towards one token per frame at the final LLM layer. Our key insight is that heuristic-based compression, widely adopted by previous methods, is prone to information loss, and this necessitates supervising LLM layers into learnable and progressive modules for token-level compression (LP-Comp). Such compression enables our VLM to digest 2x-4x more frames with improved performance. To further increase the token efficiency, we investigate frame-level compression, which selects the frames most relevant to the queries via the internal attention scores of the LLM layers, named question-conditioned compression (QC-Comp). As a notable distinction from previous studies, we mitigate the position bias of LLM attention in long contexts, i.e., the over-concentration on the beginning and end of a sequence, by splitting long videos into short segments and employing local attention. Collectively, our combined token-level and frame-level leads to an extreme compression model for long video understanding, named XComp, achieving a significantly larger compression ratio and enabling denser frame sampling. Our XComp is finetuned from VideoChat-Flash with a data-efficient supervised compression tuning stage that only requires 2.5% of the supervised fine-tuning data, yet boosts the accuracy from 42.9% to 46.2% on LVBench and enhances multiple other long video benchmarks.

Xiang Hao, Jingxiang Chen, Shixing Chen et al.

To help customers make better-informed viewing choices, video-streaming services try to moderate their content and provide more visibility into which portions of their movies and TV episodes contain age-appropriate material (e.g., nudity, sex, violence, or drug-use). Supervised models to localize these sensitive activities require large amounts of clip-level labeled data which is hard to obtain, while weakly-supervised models to this end usually do not offer competitive accuracy. To address this challenge, we propose a novel Coarse2Fine network designed to make use of readily obtainable video-level weak labels in conjunction with sparse clip-level labels of age-appropriate activities. Our model aggregates frame-level predictions to make video-level classifications and is therefore able to leverage sparse clip-level labels along with video-level labels. Furthermore, by performing frame-level predictions in a hierarchical manner, our approach is able to overcome the label-imbalance problem caused due to the rare-occurrence nature of age-appropriate content. We present comparative results of our approach using 41,234 movies and TV episodes (~3 years of video-content) from 521 sub-genres and 250 countries making it by far the largest-scale empirical analysis of age-appropriate activity localization in long-form videos ever published. Our approach offers 107.2% relative mAP improvement (from 5.5% to 11.4%) over existing state-of-the-art activity-localization approaches.

Guofeng Cui, Yang Liu, Pichao Wang et al.

Batch active learning (BAL) is a crucial technique for reducing labeling costs and improving data efficiency in training large-scale deep learning models. Traditional BAL methods often rely on metrics like Mahalanobis Distance to balance uncertainty and diversity when selecting data for annotation. However, these methods predominantly focus on the distribution of unlabeled data and fail to leverage feedback from labeled data or the model's performance. To address these limitations, we introduce TrustSet, a novel approach that selects the most informative data from the labeled dataset, ensuring a balanced class distribution to mitigate the long-tail problem. Unlike CoreSet, which focuses on maintaining the overall data distribution, TrustSet optimizes the model's performance by pruning redundant data and using label information to refine the selection process. To extend the benefits of TrustSet to the unlabeled pool, we propose a reinforcement learning (RL)-based sampling policy that approximates the selection of high-quality TrustSet candidates from the unlabeled data. Combining TrustSet and RL, we introduce the Batch Reinforcement Active Learning with TrustSet (BRAL-T) framework. BRAL-T achieves state-of-the-art results across 10 image classification benchmarks and 2 active fine-tuning tasks, demonstrating its effectiveness and efficiency in various domains.

Xiang Hao, Shixue Wen, Xiangdong Su et al.

In single-channel speech enhancement, methods based on full-band spectral features have been widely studied. However, only a few methods pay attention to non-full-band spectral features. In this paper, we explore a knowledge distillation framework based on sub-band spectral mapping for single-channel speech enhancement. Specifically, we divide the full frequency band into multiple sub-bands and pre-train an elite-level sub-band enhancement model (teacher model) for each sub-band. These teacher models are dedicated to processing their own sub-bands. Next, under the teacher models' guidance, we train a general sub-band enhancement model (student model) that works for all sub-bands. Without increasing the number of model parameters and computational complexity, the student model's performance is further improved. To evaluate our proposed method, we conducted a large number of experiments on an open-source data set. The final experimental results show that the guidance from the elite-level teacher models dramatically improves the student model's performance, which exceeds the full-band model by employing fewer parameters.

Jincao Yao, Yunpeng Wang, Zhikai Lei et al.

An artificial intelligence-generated content-enhanced computer-aided diagnosis (AIGC-CAD) model, designated as ThyGPT, has been developed. This model, inspired by the architecture of ChatGPT, could assist radiologists in assessing the risk of thyroid nodules through semantic-level human-machine interaction. A dataset comprising 19,165 thyroid nodule ultrasound cases from Zhejiang Cancer Hospital was assembled to facilitate the training and validation of the model. After training, ThyGPT could automatically evaluate thyroid nodule and engage in effective communication with physicians through human-computer interaction. The performance of ThyGPT was rigorously quantified using established metrics such as the receiver operating characteristic (ROC) curve, area under the curve (AUC), sensitivity, and specificity. The empirical findings revealed that radiologists, when supplemented with ThyGPT, markedly surpassed the diagnostic acumen of their peers utilizing traditional methods as well as the performance of the model in isolation. These findings suggest that AIGC-CAD systems, exemplified by ThyGPT, hold the promise to fundamentally transform the diagnostic workflows of radiologists in forthcoming years.

Hanglei Zhang, Yiwei Guo, Zhihan Li et al.

Most neural speech codecs achieve bitrate adjustment through intra-frame mechanisms, such as codebook dropout, at a Constant Frame Rate (CFR). However, speech segments inherently have time-varying information density (e.g., silent intervals versus voiced regions). This property makes CFR not optimal in terms of bitrate and token sequence length, hindering efficiency in real-time applications. In this work, we propose a Temporally Flexible Coding (TFC) technique, introducing variable frame rate (VFR) into neural speech codecs for the first time. TFC enables seamlessly tunable average frame rates and dynamically allocates frame rates based on temporal entropy. Experimental results show that a codec with TFC achieves optimal reconstruction quality with high flexibility, and maintains competitive performance even at lower frame rates. Our approach is promising for the integration with other efforts to develop low-frame-rate neural speech codecs for more efficient downstream tasks.

Seon-Ho Lee, Jue Wang, David Fan et al. · amazon-science

Audio Description (AD) plays a pivotal role as an application system aimed at guaranteeing accessibility in multimedia content, which provides additional narrations at suitable intervals to describe visual elements, catering specifically to the needs of visually impaired audiences. In this paper, we introduce $\mathrm{CA^3D}$, the pioneering unified Context-Aware Automatic Audio Description system that provides AD event scripts with precise locations in the long cinematic content. Specifically, $\mathrm{CA^3D}$ system consists of: 1) a Temporal Feature Enhancement Module to efficiently capture longer term dependencies, 2) an anchor-based AD event detector with feature suppression module that localizes the AD events and extracts discriminative feature for AD generation, and 3) a self-refinement module that leverages the generated output to tweak AD event boundaries from coarse to fine. Unlike conventional methods which rely on metadata and ground truth AD timestamp for AD detection and generation tasks, the proposed $\mathrm{CA^3D}$ is the first end-to-end trainable system that only uses visual cue. Extensive experiments demonstrate that the proposed $\mathrm{CA^3D}$ improves existing architectures for both AD event detection and script generation metrics, establishing the new state-of-the-art performances in the AD automation.

Yicheng Wang, Zhikang Zhang, Jue Wang et al. · amazon-science

In various video-language learning tasks, the challenge of achieving cross-modality alignment with multi-grained data persists. We propose a method to tackle this challenge from two crucial perspectives: data and modeling. Given the absence of a multi-grained video-text pretraining dataset, we introduce a Granularity EXpansion (GEX) method with Integration and Compression operations to expand the granularity of a single-grained dataset. To better model multi-grained data, we introduce an Iterative Approximation Module (IAM), which embeds multi-grained videos and texts into a unified, low-dimensional semantic space while preserving essential information for cross-modal alignment. Furthermore, GEXIA is highly scalable with no restrictions on the number of video-text granularities for alignment. We evaluate our work on three categories of video tasks across seven benchmark datasets, showcasing state-of-the-art or comparable performance. Remarkably, our model excels in tasks involving long-form video understanding, even though the pretraining dataset only contains short video clips.

Xiang Fan, Xiaohang Sun, Kushan Thakkar et al.

Effectively handling temporal redundancy remains a key challenge in learning video models. Prevailing approaches often treat each set of frames independently, failing to effectively capture the temporal dependencies and redundancies inherent in videos. To address this limitation, we introduce RefTok, a novel reference-based tokenization method capable of capturing complex temporal dynamics and contextual information. Our method encodes and decodes sets of frames conditioned on an unquantized reference frame. When decoded, RefTok preserves the continuity of motion and the appearance of objects across frames. For example, RefTok retains facial details despite head motion, reconstructs text correctly, preserves small patterns, and maintains the legibility of handwriting from the context. Across 4 video datasets (K600, UCF-101, BAIR Robot Pushing, and DAVIS), RefTok significantly outperforms current state-of-the-art tokenizers (Cosmos and MAGVIT) and improves all evaluated metrics (PSNR, SSIM, LPIPS) by an average of 36.7% at the same or higher compression ratios. When a video generation model is trained using RefTok's latents on the BAIR Robot Pushing task, the generations not only outperform MAGVIT-B but the larger MAGVIT-L, which has 4x more parameters, across all generation metrics by an average of 27.9%.

Di Xu, Yang Zhao, Xiang Hao et al.

We introduce a novel dataset consisting of images depicting pink eggs that have been identified as Pomacea canaliculata eggs, accompanied by corresponding bounding box annotations. The purpose of this dataset is to aid researchers in the analysis of the spread of Pomacea canaliculata species by utilizing deep learning techniques, as well as supporting other investigative pursuits that require visual data pertaining to the eggs of Pomacea canaliculata. It is worth noting, however, that the identity of the eggs in question is not definitively established, as other species within the same taxonomic family have been observed to lay similar-looking eggs in regions of the Americas. Therefore, a crucial prerequisite to any decision regarding the elimination of these eggs would be to establish with certainty whether they are exclusively attributable to invasive Pomacea canaliculata or if other species are also involved. The dataset is available at https://www.kaggle.com/datasets/deeshenzhen/pinkeggs

Zhenhao Jin, Xiang Hao, Xiangdong Su

The vast majority of speech separation methods assume that the number of speakers is known in advance, hence they are specific to the number of speakers. By contrast, a more realistic and challenging task is to separate a mixture in which the number of speakers is unknown. This paper formulates the speech separation with the unknown number of speakers as a multi-pass source extraction problem and proposes a coarse-to-fine recursive speech separation method. This method comprises two stages, namely, recursive cue extraction and target speaker extraction. The recursive cue extraction stage determines how many computational iterations need to be performed and outputs a coarse cue speech by monitoring statistics in the mixture. As the number of recursive iterations increases, the accumulation of distortion eventually comes into the extracted speech and reminder. Therefore, in the second stage, we use a target speaker extraction network to extract a fine speech based on the coarse target cue and the original distortionless mixture. Experiments show that the proposed method archived state-of-the-art performance on the WSJ0 dataset with a different number of speakers. Furthermore, it generalizes well to an unseen large number of speakers.

Shixing Chen, Chun-Hao Liu, Xiang Hao et al.

Understanding scenes in movies is crucial for a variety of applications such as video moderation, search, and recommendation. However, labeling individual scenes is a time-consuming process. In contrast, movie level metadata (e.g., genre, synopsis, etc.) regularly gets produced as part of the film production process, and is therefore significantly more commonly available. In this work, we propose a novel contrastive learning approach that uses movie metadata to learn a general-purpose scene representation. Specifically, we use movie metadata to define a measure of movie similarity, and use it during contrastive learning to limit our search for positive scene-pairs to only the movies that are considered similar to each other. Our learned scene representation consistently outperforms existing state-of-the-art methods on a diverse set of tasks evaluated using multiple benchmark datasets. Notably, our learned representation offers an average improvement of 7.9% on the seven classification tasks and 9.7% improvement on the two regression tasks in LVU dataset. Furthermore, using a newly collected movie dataset, we present comparative results of our scene representation on a set of video moderation tasks to demonstrate its generalizability on previously less explored tasks.

Xiang Hao, Xiangdong Su, Zhiyu Wang et al.

Speech enhancement at extremely low signal-to-noise ratio (SNR) condition is a very challenging problem and rarely investigated in previous works. This paper proposes a robust speech enhancement approach (UNetGAN) based on U-Net and generative adversarial learning to deal with this problem. This approach consists of a generator network and a discriminator network, which operate directly in the time domain. The generator network adopts a U-Net like structure and employs dilated convolution in the bottleneck of it. We evaluate the performance of the UNetGAN at low SNR conditions (up to -20dB) on the public benchmark. The result demonstrates that it significantly improves the speech quality and substantially outperforms the representative deep learning models, including SEGAN, cGAN fo SE, Bidirectional LSTM using phase-sensitive spectrum approximation cost function (PSA-BLSTM) and Wave-U-Net regarding Short-Time Objective Intelligibility (STOI) and Perceptual evaluation of speech quality (PESQ).

Xiang Hao, Xiangdong Su, Radu Horaud et al.

This paper proposes a full-band and sub-band fusion model, named as FullSubNet, for single-channel real-time speech enhancement. Full-band and sub-band refer to the models that input full-band and sub-band noisy spectral feature, output full-band and sub-band speech target, respectively. The sub-band model processes each frequency independently. Its input consists of one frequency and several context frequencies. The output is the prediction of the clean speech target for the corresponding frequency. These two types of models have distinct characteristics. The full-band model can capture the global spectral context and the long-distance cross-band dependencies. However, it lacks the ability to modeling signal stationarity and attending the local spectral pattern. The sub-band model is just the opposite. In our proposed FullSubNet, we connect a pure full-band model and a pure sub-band model sequentially and use practical joint training to integrate these two types of models' advantages. We conducted experiments on the DNS challenge (INTERSPEECH 2020) dataset to evaluate the proposed method. Experimental results show that full-band and sub-band information are complementary, and the FullSubNet can effectively integrate them. Besides, the performance of the FullSubNet also exceeds that of the top-ranked methods in the DNS Challenge (INTERSPEECH 2020).

Huali Xu, Xiangdong Su, Meng Wang et al.

In the image inpainting task, the ability to repair both high-frequency and low-frequency information in the missing regions has a substantial influence on the quality of the restored image. However, existing inpainting methods usually fail to consider both high-frequency and low-frequency information simultaneously. To solve this problem, this paper proposes edge information and mask shrinking based image inpainting approach, which consists of two models. The first model is an edge generation model used to generate complete edge information from the damaged image, and the second model is an image completion model used to fix the missing regions with the generated edge information and the valid contents of the damaged image. The mask shrinking strategy is employed in the image completion model to track the areas to be repaired. The proposed approach is evaluated qualitatively and quantitatively on the dataset Places2. The result shows our approach outperforms state-of-the-art methods.

Xiang Hao, Xiangdong Su, Zhiyu Wang et al.

It is very challenging for speech enhancement methods to achieves robust performance under both high signal-to-noise ratio (SNR) and low SNR simultaneously. In this paper, we propose a method that integrates an SNR-based teachers-student technique and time-domain U-Net to deal with this problem. Specifically, this method consists of multiple teacher models and a student model. We first train the teacher models under multiple small-range SNRs that do not coincide with each other so that they can perform speech enhancement well within the specific SNR range. Then, we choose different teacher models to supervise the training of the student model according to the SNR of the training data. Eventually, the student model can perform speech enhancement under both high SNR and low SNR. To evaluate the proposed method, we constructed a dataset with an SNR ranging from -20dB to 20dB based on the public dataset. We experimentally analyzed the effectiveness of the SNR-based teachers-student technique and compared the proposed method with several state-of-the-art methods.