Search

Removing audio effects from electric guitar recordings makes it easier for post-production and sound editing. An audio distortion recovery model not only improves the clarity of the guitar sounds but also opens up new opportunities for creative adjustments in mixing and mastering. While progress have been made in creating such models, previous efforts have largely focused on synthetic distortions that may be too simplistic to accurately capture the complexities seen in real-world recordings. In this paper, we tackle the task by using a dataset of guitar recordings rendered with commercial-grade audio effect VST plugins. Moreover, we introduce a novel two-stage methodology for audio distortion recovery. The idea is to firstly process the audio signal in the Mel-spectrogram domain in the first stage, and then use a neural vocoder to generate the pristine original guitar sound from the processed Mel-spectrogram in the second stage. We report a set of experiments demonstrating the effectiveness of our approach over existing methods, through both subjective and objective evaluation metrics., Comment: DAFx 2024

Change captioning aims to succinctly describe the semantic change between a pair of similar images, while being immune to distractors (illumination and viewpoint changes). Under these distractors, unchanged objects often appear pseudo changes about location and scale, and certain objects might overlap others, resulting in perturbational and discrimination-degraded features between two images. However, most existing methods directly capture the difference between them, which risk obtaining error-prone difference features. In this paper, we propose a distractors-immune representation learning network that correlates the corresponding channels of two image representations and decorrelates different ones in a self-supervised manner, thus attaining a pair of stable image representations under distractors. Then, the model can better interact them to capture the reliable difference features for caption generation. To yield words based on the most related difference features, we further design a cross-modal contrastive regularization, which regularizes the cross-modal alignment by maximizing the contrastive alignment between the attended difference features and generated words. Extensive experiments show that our method outperforms the state-of-the-art methods on four public datasets. The code is available at https://github.com/tuyunbin/DIRL., Comment: Accepted by ECCV 2024

Sequential trial emulation (STE) is an approach to estimating causal treatment effects by emulating a sequence of target trials from observational data. In STE, inverse probability weighting is commonly utilised to address time-varying confounding and/or dependent censoring. Then structural models for potential outcomes are applied to the weighted data to estimate treatment effects. For inference, the simple sandwich variance estimator is popular but conservative, while nonparametric bootstrap is computationally expensive, and a more efficient alternative, linearised estimating function (LEF) bootstrap, has not been adapted to STE. We evaluated the performance of various methods for constructing confidence intervals (CIs) of marginal risk differences in STE with survival outcomes by comparing the coverage of CIs based on nonparametric/LEF bootstrap, jackknife, and the sandwich variance estimator through simulations. LEF bootstrap CIs demonstrated the best coverage with small/moderate sample sizes, low event rates and low treatment prevalence, which were the motivating scenarios for STE. They were less affected by treatment group imbalance and faster to compute than nonparametric bootstrap CIs. With large sample sizes and medium/high event rates, the sandwich-variance-estimator-based CIs had the best coverage and were the fastest to compute. These findings offer guidance in constructing CIs in causal survival analysis using STE., Comment: Main text: 23 pages, 5 figures, 5 tables. Supplementary materials included

Expressive music synthesis (EMS) for violin performance is a challenging task due to the disagreement among music performers in the interpretation of expressive musical terms (EMTs), scarcity of labeled recordings, and limited generalization ability of the synthesis model. These challenges create trade-offs between model effectiveness, diversity of generated results, and controllability of the synthesis system, making it essential to conduct a comparative study on EMS model design. This paper explores two violin EMS approaches. The end-to-end approach is a modification of a state-of-the-art text-to-speech generator. The parameter-controlled approach is based on a simple parameter sampling process that can render note lengths and other parameters compatible with MIDI-DDSP. We study these two approaches (in total, three model variants) through objective and subjective experiments and discuss several key issues of EMS based on the results., Comment: 15 pages, 2 figures, 3 tables

In cross-modal music processing, translation between visual, auditory, and semantic content opens up new possibilities as well as challenges. The construction of such a transformative scheme depends upon a benchmark corpus with a comprehensive data infrastructure. In particular, the assembly of a large-scale cross-modal dataset presents major challenges. In this paper, we present the MOSA (Music mOtion with Semantic Annotation) dataset, which contains high quality 3-D motion capture data, aligned audio recordings, and note-by-note semantic annotations of pitch, beat, phrase, dynamic, articulation, and harmony for 742 professional music performances by 23 professional musicians, comprising more than 30 hours and 570 K notes of data. To our knowledge, this is the largest cross-modal music dataset with note-level annotations to date. To demonstrate the usage of the MOSA dataset, we present several innovative cross-modal music information retrieval (MIR) and musical content generation tasks, including the detection of beats, downbeats, phrase, and expressive contents from audio, video and motion data, and the generation of musicians' body motion from given music audio. The dataset and codes are available alongside this publication (https://github.com/yufenhuang/MOSA-Music-mOtion-and-Semantic-Annotation-dataset)., Comment: IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2024. 14 pages, 7 figures. Dataset is available on: https://github.com/yufenhuang/MOSA-Music-mOtion-and-Semantic-Annotation-dataset/tree/main and https://zenodo.org/records/11393449

Multi-change captioning aims to describe complex and coupled changes within an image pair in natural language. Compared with single-change captioning, this task requires the model to have higher-level cognition ability to reason an arbitrary number of changes. In this paper, we propose a novel context-aware difference distilling (CARD) network to capture all genuine changes for yielding sentences. Given an image pair, CARD first decouples context features that aggregate all similar/dissimilar semantics, termed common/difference context features. Then, the consistency and independence constraints are designed to guarantee the alignment/discrepancy of common/difference context features. Further, the common context features guide the model to mine locally unchanged features, which are subtracted from the pair to distill locally difference features. Next, the difference context features augment the locally difference features to ensure that all changes are distilled. In this way, we obtain an omni-representation of all changes, which is translated into linguistic sentences by a transformer decoder. Extensive experiments on three public datasets show CARD performs favourably against state-of-the-art methods.The code is available at https://github.com/tuyunbin/CARD., Comment: Accepted by ACL 2024 main conference (long paper)

Randomised controlled trials (RCTs) are regarded as the gold standard for estimating causal treatment effects on health outcomes. However, RCTs are not always feasible, because of time, budget or ethical constraints. Observational data such as those from electronic health records (EHRs) offer an alternative way to estimate the causal effects of treatments. Recently, the `target trial emulation' framework was proposed by Hernan and Robins (2016) to provide a formal structure for estimating causal treatment effects from observational data. To promote more widespread implementation of target trial emulation in practice, we develop the R package TrialEmulation to emulate a sequence of target trials using observational time-to-event data, where individuals who start to receive treatment and those who have not been on the treatment at the baseline of the emulated trials are compared in terms of their risks of an outcome event. Specifically, TrialEmulation provides (1) data preparation for emulating a sequence of target trials, (2) calculation of the inverse probability of treatment and censoring weights to handle treatment switching and dependent censoring, (3) fitting of marginal structural models for the time-to-event outcome given baseline covariates, (4) estimation and inference of marginal intention to treat and per-protocol effects of the treatment in terms of marginal risk differences between treated and untreated for a user-specified target trial population. In particular, TrialEmulation can accommodate large data sets (e.g., from EHRs) within memory constraints of R by processing data in chunks and applying case-control sampling. We demonstrate the functionality of TrialEmulation using a simulated data set that mimics typical observational time-to-event data in practice., Comment: 42 pages, 4 figures, 5 tables

Many deep learning models have achieved dominant performance on the offline beat tracking task. However, online beat tracking, in which only the past and present input features are available, still remains challenging. In this paper, we propose BEAt tracking Streaming Transformer (BEAST), an online joint beat and downbeat tracking system based on the streaming Transformer. To deal with online scenarios, BEAST applies contextual block processing in the Transformer encoder. Moreover, we adopt relative positional encoding in the attention layer of the streaming Transformer encoder to capture relative timing position which is critically important information in music. Carrying out beat and downbeat experiments on benchmark datasets for a low latency scenario with maximum latency under 50 ms, BEAST achieves an F1-measure of 80.04% in beat and 46.78% in downbeat, which is a substantial improvement of about 5 percentage points over the state-of-the-art online beat tracking model., Comment: Accepted by ICASSP 2024

Graph computing has become increasingly crucial in processing large-scale graph data, with numerous systems developed for this purpose. Two years ago, we introduced GraphScope as a system addressing a wide array of graph computing needs, including graph traversal, analytics, and learning in one system. Since its inception, GraphScope has achieved significant technological advancements and gained widespread adoption across various industries. However, one key lesson from this journey has been understanding the limitations of a "one-size-fits-all" approach, especially when dealing with the diversity of programming interfaces, applications, and data storage formats in graph computing. In response to these challenges, we present GraphScope Flex, the next iteration of GraphScope. GraphScope Flex is designed to be both resource-efficient and cost-effective, while also providing flexibility and user-friendliness through its LEGO-like modularity. This paper explores the architectural innovations and fundamental design principles of GraphScope Flex, all of which are direct outcomes of the lessons learned during our ongoing development process. We validate the adaptability and efficiency of GraphScope Flex with extensive evaluations on synthetic and real-world datasets. The results show that GraphScope Flex achieves 2.4X throughput and up to 55.7X speedup over other systems on the LDBC Social Network and Graphalytics benchmarks, respectively. Furthermore, GraphScope Flex accomplishes up to a 2,400X performance gain in real-world applications, demonstrating its proficiency across a wide range of graph computing scenarios with increased effectiveness.

Dynamic treatment regimes (DTRs) formalize medical decision-making as a sequence of rules for different stages, mapping patient-level information to recommended treatments. In practice, estimating an optimal DTR using observational data from electronic medical record (EMR) databases can be complicated by nonignorable missing covariates resulting from informative monitoring of patients. Since complete case analysis can provide consistent estimation of outcome model parameters under the assumption of outcome-independent missingness, Q-learning is a natural approach to accommodating nonignorable missing covariates. However, the backward induction algorithm used in Q-learning can introduce challenges, as nonignorable missing covariates at later stages can result in nonignorable missing pseudo-outcomes at earlier stages, leading to suboptimal DTRs, even if the longitudinal outcome variables are fully observed. To address this unique missing data problem in DTR settings, we propose two weighted Q-learning approaches where inverse probability weights for missingness of the pseudo-outcomes are obtained through estimating equations with valid nonresponse instrumental variables or sensitivity analysis. The asymptotic properties of the weighted Q-learning estimators are derived, and the finite-sample performance of the proposed methods is evaluated and compared with alternative methods through extensive simulation studies. Using EMR data from the Medical Information Mart for Intensive Care database, we apply the proposed methods to investigate the optimal fluid strategy for sepsis patients in intensive care units., Comment: This paper includes a supplementary material in the latter part of this arXiv version

The tasks of automatic lyrics transcription and lyrics alignment have witnessed significant performance improvements in the past few years. However, most of the previous works only focus on English in which large-scale datasets are available. In this paper, we address lyrics transcription and alignment of polyphonic Mandarin pop music in a low-resource setting. To deal with the data scarcity issue, we adapt pretrained Whisper model and fine-tune it on a monophonic Mandarin singing dataset. With the use of data augmentation and source separation model, results show that the proposed method achieves a character error rate of less than 18% on a Mandarin polyphonic dataset for lyrics transcription, and a mean absolute error of 0.071 seconds for lyrics alignment. Our results demonstrate the potential of adapting a pretrained speech model for lyrics transcription and alignment in low-resource scenarios., Comment: Accepted by ASRU 2023

Livestreaming commerce, a hybrid of e-commerce and self-media, has expanded the broad spectrum of traditional sales performance determinants. To investigate the factors that contribute to the success of livestreaming commerce, we construct a longitudinal firm-level database with 19,175 observations, covering an entire livestreaming subsector. By comparing the forecasting accuracy of eight machine learning models, we identify a random forest model that provides the best prediction of gross merchandise volume (GMV). Furthermore, we utilize explainable artificial intelligence to open the black-box of machine learning model, discovering four new facts: 1) variables representing the popularity of livestreaming events are crucial features in predicting GMV. And voice attributes are more important than appearance; 2) popularity is a major determinant of sales for female hosts, while vocal aesthetics is more decisive for their male counterparts; 3) merits and drawbacks of the voice are not equally valued in the livestreaming market; 4) based on changes of comments, page views and likes, sales growth can be divided into three stages. Finally, we innovatively propose a 3D-SHAP diagram that demonstrates the relationship between predicting feature importance, target variable, and its predictors. This diagram identifies bottlenecks for both beginner and top livestreamers, providing insights into ways to optimize their sales performance., Comment: 25 pages, 10 figures

Objective: Motor Imagery (MI) serves as a crucial experimental paradigm within the realm of Brain Computer Interfaces (BCIs), aiming to decoding motor intentions from electroencephalogram (EEG) signals. Method: Drawing inspiration from Riemannian geometry and Cross-Frequency Coupling (CFC), this paper introduces a novel approach termed Riemann Tangent Space Mapping using Dichotomous Filter Bank with Convolutional Neural Network (DFBRTS) to enhance the representation quality and decoding capability pertaining to MI features. DFBRTS first initiates the process by meticulously filtering EEG signals through a Dichotomous Filter Bank, structured in the fashion of a complete binary tree. Subsequently, it employs Riemann Tangent Space Mapping to extract salient EEG signal features within each sub-band. Finally, a lightweight convolutional neural network is employed for further feature extraction and classification, operating under the joint supervision of cross-entropy and center loss. To validate the efficacy, extensive experiments were conducted using DFBRTS on two well-established benchmark datasets: the BCI competition IV 2a (BCIC-IV-2a) dataset and the OpenBMI dataset. The performance of DFBRTS was benchmarked against several state-of-the-art MI decoding methods, alongside other Riemannian geometry-based MI decoding approaches. Results: DFBRTS significantly outperforms other MI decoding algorithms on both datasets, achieving a remarkable classification accuracy of 78.16% for four-class and 71.58% for two-class hold-out classification, as compared to the existing benchmarks., Comment: 22 pages, 7 figures

Change captioning aims to describe the difference between a pair of similar images. Its key challenge is how to learn a stable difference representation under pseudo changes caused by viewpoint change. In this paper, we address this by proposing a self-supervised cross-view representation reconstruction (SCORER) network. Concretely, we first design a multi-head token-wise matching to model relationships between cross-view features from similar/dissimilar images. Then, by maximizing cross-view contrastive alignment of two similar images, SCORER learns two view-invariant image representations in a self-supervised way. Based on these, we reconstruct the representations of unchanged objects by cross-attention, thus learning a stable difference representation for caption generation. Further, we devise a cross-modal backward reasoning to improve the quality of caption. This module reversely models a ``hallucination'' representation with the caption and ``before'' representation. By pushing it closer to the ``after'' representation, we enforce the caption to be informative about the difference in a self-supervised manner. Extensive experiments show our method achieves the state-of-the-art results on four datasets. The code is available at https://github.com/tuyunbin/SCORER., Comment: Accepted by ICCV 2023

A hypertree is a connected hypergraph without cycles. Further a hypertree is called an $r$-tree if, additionally, it is $r$-uniform. Note that 2-trees are just ordinary trees. A classical result states that for any 2-tree $T$ with characteristic polynomial $\phi_T(\lambda)$ and matching polynomial $\varphi_T(\lambda)$, then $\phi_T(\lambda)=\varphi_T(\lambda).$ More generally, suppose $\mathcal{T}$ is an $r$-tree of size $m$ with $r\geq2$. In this paper, we extend the above classical relationship to $r$-trees and establish that \[ \phi_{\mathcal{T}}(\lambda)=\prod_{H \sqsubseteq \mathcal{T}}\varphi_{H}(\lambda)^{a_{H}}, \] where the product is over all connected subgraphs $H$ of $\mathcal{T}$, and the exponent $a_{H}$ of the factor $\varphi_{H}(\lambda)$ can be written as \[ a_H=b^{m-e(H)-|\partial(H)|}c^{e(H)}(b-c)^{|\partial(H)|}, \] where $e(H)$ is the size of $H$, $\partial(H)$ is the boundary of $H$, and $b=(r-1)^{r-1}, c=r^{r-2}$. In particular, for $r=2$, the above correspondence reduces to the classical result for ordinary trees. In addition, we resolve a conjecture by Clark-Cooper [{\em Electron. J. Combin.}, 2018] and show that for any subgraph $H$ of an $r$-tree $\mathcal{T}$ with $r\geq3$, $\varphi_H(\lambda)$ divides $\phi_{\mathcal{T}}(\lambda)$, and additionally $\phi_H(\lambda)$ divides $\phi_{\mathcal{T}}(\lambda)$, if either $r\geq 4$ or $H$ is connected when $r=3$. Moreover, a counterexample is given for the case when $H$ is a disconnected subgraph of a 3-tree., Comment: 36 pages, 4 figures