Xulong Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xulong Zhang has a total of 66 co-authors. Here are the top co-authors by number of shared publications:
- Jianzong Wang
Ping An Technology (Shenzhen) Co., Ltd., Shenzhen, Guangdong, China
14 shared publications Jianzong Wang
Ping An Technology (Shenzhen) Co., Ltd
12 shared publicationsJing Xiao
Ping An Technology (Shenzhen) Co., Ltd
11 shared publications- Jing Xiao
Pingan Group Co., Ltd., Shenzhen, China
10 shared publications Ning Cheng
Ping An Technology (Shenzhen) Co., Ltd
10 shared publications- Wei Li
School of Computer Science and Technology, Fudan University, Shanghai 200433, China
9 shared publications
Latest Publications
- 22 November 2025
- Published by Springer Nature
- p. 91-105
Turbo-TTS: Enhancing Diffusion Model TTS with an Improved ODE Solver
- 22 November 2025
- Published by Springer Nature
- p. 91-105
Turbo-TTS: Enhancing Diffusion Model TTS with an Improved ODE Solver
- 1 October 2025
- Published by Zhejiang University Press in Frontiers of Information Technology & Electronic Engineering
- Vol. 26 (10) , 1793-1808
Knowledge distillation for financial large language models: a systematic review of strategies, applications, and evaluation
Financial large language models (FinLLMs) offer immense potential for financial applications. While excessive deployment expenditures and considerable infe
- 1 October 2025
- Published by Zhejiang University Press in Frontiers of Information Technology & Electronic Engineering
- Vol. 26 (10) , 1793-1808
Knowledge distillation for financial large language models: a systematic review of strategies, applications, and evaluation
Financial large language models (FinLLMs) offer immense potential for financial applications. While excessive deployment expenditures and considerable infe
- Open Access
IDEAW: Robust Neural Audio Watermarking with Invertible Dual-Embedding
The audio watermarking technique embeds messages into audio and accurately extracts messages from the watermarked audio. Traditional methods develop algorithms based on expert experience to embed watermarks into the time-domain or transform-domain of signals. With the development of deep neural networks, deep learning-based neural audio watermarking has emerged. Compared to traditional algorithms, neural audio watermarking achieves better robustness by considering various attacks during training. However, current neural watermarking methods suffer from low capacity and unsatisfactory imperceptibility. Additionally, the issue of watermark locating, which is extremely important and even more pronounced in neural audio watermarking, has not been adequately studied. In this paper, we design a dual-embedding watermarking model for efficient locating. We also consider the impact of the attack layer on the invertible neural network in robustness training, improving the model to enhance both its reasonableness and stability. Experiments show that the proposed model, IDEAW, can withstand various attacks with higher capacity and more efficient locating ability compared to existing methods.
Open AccessIDEAW: Robust Neural Audio Watermarking with Invertible Dual-Embedding
The audio watermarking technique embeds messages into audio and accurately extracts messages from the watermarked audio. Traditional methods develop algorithms based on expert experience to embed watermarks into the time-domain or transform-domain of signals. With the development of deep neural networks, deep learning-based neural audio watermarking has emerged. Compared to traditional algorithms, neural audio watermarking achieves better robustness by considering various attacks during training. However, current neural watermarking methods suffer from low capacity and unsatisfactory imperceptibility. Additionally, the issue of watermark locating, which is extremely important and even more pronounced in neural audio watermarking, has not been adequately studied. In this paper, we design a dual-embedding watermarking model for efficient locating. We also consider the impact of the attack layer on the invertible neural network in robustness training, improving the model to enhance both its reasonableness and stability. Experiments show that the proposed model, IDEAW, can withstand various attacks with higher capacity and more efficient locating ability compared to existing methods.
- 28 August 2024
- Published by Springer Nature
- p. 90-104
RSET: Remapping-Based Sorting Method for Emotion Transfer Speech Synthesis
- 28 August 2024
- Published by Springer Nature
- p. 90-104
RSET: Remapping-Based Sorting Method for Emotion Transfer Speech Synthesis
- Open Access
- 14 August 2024
- Published by Bio-Byword Scientific Publishing, Pty. Ltd. in Journal of Electronic Research and Application
- Vol. 8 (5) , 125-133
A Novel Optimization Scheme for Named Entity Recognition with Pre-trained Language Models
Named Entity Recognition (NER) is crucial for extracting structured information from text. While traditional methods rely on rules, Conditional Random Fields (CRFs), or deep learning, the advent of large-scale Pre-trained Language Models (PLMs) offers new possibilities. PLMs excel at contextual learning, potentially simplifying many natural language processing tasks. However, their application to NER remains underexplored. This paper investigates leveraging the GPT-3 PLM for NER without fine-tuning. We propose a novel scheme that utilizes carefully crafted templates and context examples selected based on semantic similarity. Our experimental results demonstrate the feasibility of this approach, suggesting a promising direction for harnessing PLMs in NER.
- 14 August 2024
- Published by Bio-Byword Scientific Publishing, Pty. Ltd. in Journal of Electronic Research and Application
- Vol. 8 (5) , 125-133
Open AccessA Novel Optimization Scheme for Named Entity Recognition with Pre-trained Language Models
Named Entity Recognition (NER) is crucial for extracting structured information from text. While traditional methods rely on rules, Conditional Random Fields (CRFs), or deep learning, the advent of large-scale Pre-trained Language Models (PLMs) offers new possibilities. PLMs excel at contextual learning, potentially simplifying many natural language processing tasks. However, their application to NER remains underexplored. This paper investigates leveraging the GPT-3 PLM for NER without fine-tuning. We propose a novel scheme that utilizes carefully crafted templates and context examples selected based on semantic similarity. Our experimental results demonstrate the feasibility of this approach, suggesting a promising direction for harnessing PLMs in NER.
- 1 August 2024
- Published by Springer Nature
- p. 374-385
RREH: Reconstruction Relations Embedded Hashing for Semi-paired Cross-Modal Retrieval
- 1 August 2024
- Published by Springer Nature
- p. 374-385
RREH: Reconstruction Relations Embedded Hashing for Semi-paired Cross-Modal Retrieval
- Open Access
Learning Expressive Disentangled Speech Representations with Soft Speech Units and Adversarial Style Augmentation
Voice conversion is the task to transform voice characteristics of source speech while preserving content information. Nowadays, self-supervised representation learning models are increasingly utilized in content extraction. However, in these representations, a lot of hidden speaker information leads to timbre leakage while the prosodic information of hidden units lacks use. To address these issues, we propose a novel framework for expressive voice conversion called "SAVC" based on soft speech units from HuBert-soft. Taking soft speech units as input, we design an attribute encoder to extract content and prosody features respectively. Specifically, we first introduce statistic perturbation imposed by adversarial style augmentation to eliminate speaker information. Then the prosody is implicitly modeled on soft speech units with knowledge distillation. Experiment results show that the intelligibility and naturalness of converted speech outperform previous work.
Open AccessLearning Expressive Disentangled Speech Representations with Soft Speech Units and Adversarial Style Augmentation
Voice conversion is the task to transform voice characteristics of source speech while preserving content information. Nowadays, self-supervised representation learning models are increasingly utilized in content extraction. However, in these representations, a lot of hidden speaker information leads to timbre leakage while the prosodic information of hidden units lacks use. To address these issues, we propose a novel framework for expressive voice conversion called "SAVC" based on soft speech units from HuBert-soft. Taking soft speech units as input, we design an attribute encoder to extract content and prosody features respectively. Specifically, we first introduce statistic perturbation imposed by adversarial style augmentation to eliminate speaker information. Then the prosody is implicitly modeled on soft speech units with knowledge distillation. Experiment results show that the intelligibility and naturalness of converted speech outperform previous work.
- Open Access
Medical Speech Symptoms Classification via Disentangled Representation
Intent is defined for understanding spoken language in existing works. Both textual features and acoustic features involved in medical speech contain intent, which is important for symptomatic diagnosis. In this paper, we propose a medical speech classification model named DRSC that automatically learns to disentangle intent and content representations from textual-acoustic data for classification. The intent representations of the text domain and the Mel-spectrogram domain are extracted via intent encoders, and then the reconstructed text feature and the Mel-spectrogram feature are obtained through two exchanges. After combining the intent from two domains into a joint representation, the integrated intent representation is fed into a decision layer for classification. Experimental results show that our model obtains an average accuracy rate of 95% in detecting 25 different medical symptoms.
Open AccessMedical Speech Symptoms Classification via Disentangled Representation
Intent is defined for understanding spoken language in existing works. Both textual features and acoustic features involved in medical speech contain intent, which is important for symptomatic diagnosis. In this paper, we propose a medical speech classification model named DRSC that automatically learns to disentangle intent and content representations from textual-acoustic data for classification. The intent representations of the text domain and the Mel-spectrogram domain are extracted via intent encoders, and then the reconstructed text feature and the Mel-spectrogram feature are obtained through two exchanges. After combining the intent from two domains into a joint representation, the integrated intent representation is fed into a decision layer for classification. Experimental results show that our model obtains an average accuracy rate of 95% in detecting 25 different medical symptoms.
- Open Access
QLSC: A Query Latent Semantic Calibrator for Robust Extractive Question Answering
Extractive Question Answering (EQA) in Machine Reading Comprehension (MRC) often faces the challenge of dealing with semantically identical but format-variant inputs. Our work introduces a novel approach, called the ``Query Latent Semantic Calibrator (QLSC)'', designed as an auxiliary module for existing MRC models. We propose a unique scaling strategy to capture latent semantic center features of queries. These features are then seamlessly integrated into traditional query and passage embeddings using an attention mechanism. By deepening the comprehension of the semantic queries-passage relationship, our approach diminishes sensitivity to variations in text format and boosts the model's capability in pinpointing accurate answers. Experimental results on robust Question-Answer datasets confirm that our approach effectively handles format-variant but semantically identical queries, highlighting the effectiveness and adaptability of our proposed method.
Open AccessQLSC: A Query Latent Semantic Calibrator for Robust Extractive Question Answering
Extractive Question Answering (EQA) in Machine Reading Comprehension (MRC) often faces the challenge of dealing with semantically identical but format-variant inputs. Our work introduces a novel approach, called the ``Query Latent Semantic Calibrator (QLSC)'', designed as an auxiliary module for existing MRC models. We propose a unique scaling strategy to capture latent semantic center features of queries. These features are then seamlessly integrated into traditional query and passage embeddings using an attention mechanism. By deepening the comprehension of the semantic queries-passage relationship, our approach diminishes sensitivity to variations in text format and boosts the model's capability in pinpointing accurate answers. Experimental results on robust Question-Answer datasets confirm that our approach effectively handles format-variant but semantically identical queries, highlighting the effectiveness and adaptability of our proposed method.
- Open Access
EfficientASR: Speech Recognition Network Compression via Attention Redundancy and Chunk-Level FFN Optimization
In recent years, Transformer networks have shown remarkable performance in speech recognition tasks. However, their deployment poses challenges due to high computational and storage resource requirements. To address this issue, a lightweight model called EfficientASR is proposed in this paper, aiming to enhance the versatility of Transformer models. EfficientASR employs two primary modules: Shared Residual Multi-Head Attention (SRMHA) and Chunk-Level Feedforward Networks (CFFN). The SRMHA module effectively reduces redundant computations in the network, while the CFFN module captures spatial knowledge and reduces the number of parameters. The effectiveness of the EfficientASR model is validated on two public datasets, namely Aishell-1 and HKUST. Experimental results demonstrate a 36% reduction in parameters compared to the baseline Transformer network, along with improvements of 0.3% and 0.2% in Character Error Rate (CER) on the Aishell-1 and HKUST datasets, respectively.
Open AccessEfficientASR: Speech Recognition Network Compression via Attention Redundancy and Chunk-Level FFN Optimization
In recent years, Transformer networks have shown remarkable performance in speech recognition tasks. However, their deployment poses challenges due to high computational and storage resource requirements. To address this issue, a lightweight model called EfficientASR is proposed in this paper, aiming to enhance the versatility of Transformer models. EfficientASR employs two primary modules: Shared Residual Multi-Head Attention (SRMHA) and Chunk-Level Feedforward Networks (CFFN). The SRMHA module effectively reduces redundant computations in the network, while the CFFN module captures spatial knowledge and reduces the number of parameters. The effectiveness of the EfficientASR model is validated on two public datasets, namely Aishell-1 and HKUST. Experimental results demonstrate a 36% reduction in parameters compared to the baseline Transformer network, along with improvements of 0.3% and 0.2% in Character Error Rate (CER) on the Aishell-1 and HKUST datasets, respectively.