arXiv subjects

Zhen-Hua Ling

Publications and source records attributed to Zhen-Hua Ling.

At least 55 records · Page 3Linked to original sources

arXiv2025

A Distilled Low-Latency Neural Vocoder with Explicit Amplitude and Phase Prediction

The majority of mainstream neural vocoders primarily focus on speech quality and generation speed, while overlooking latency, which is a critical factor in real-time applications. Excessive latency leads to noticeable delays in user interaction, severely degrading the user experience and rendering such systems impractical for real-time use. Therefore, this paper proposes DLL-APNet, a Distilled Low-Latency neural vocoder which first predicts the Amplitude and Phase spectra explicitly from input mel spectrogram and then reconstructs the speech waveform via inverse short-time Fourier transform (iSTFT). The DLL-APNet vocoder leverages causal convolutions to constrain the utilization of information to current and historical contexts, effectively minimizing latency. To mitigate speech quality degradation caused by causal constraints, a knowledge distillation strategy is proposed, where a pre-trained non-causal teacher vocoder guides intermediate feature generation of the causal student DLL-APNet vocoder. Experimental results demonstrate that the proposed DLL-APNet vocoder produces higher-quality speech than other causal vocoders, while requiring fewer computational resources. Furthermore, the proposed DLL-APNet vocoder achieves speech quality on par with mainstream non-causal neural vocoders, validating its ability to deliver both high perceptual quality and low latency.

eess.AS↗

arXiv2025

A High-Quality and Low-Complexity Streamable Neural Speech Codec with Knowledge Distillation

While many current neural speech codecs achieve impressive reconstructed speech quality, they often neglect latency and complexity considerations, limiting their practical deployment in downstream tasks such as real-time speech communication and efficient speech compression. In our previous work, we proposed StreamCodec, which enables streamable speech coding by leveraging model causalization and a scalar-vector-combined quantization strategy, but its reconstructed quality and complexity still have room for improvement. Therefore, this paper proposes an improved iteration of StreamCodec, named StreamCodec2. The StreamCodec2 supports streamable and lightweight speech coding by adopting a fully causal architecture and reducing the convolutional channels. To compensate for the speech quality degradation caused by model causalization and pruning, we introduce a non-causal, high-complexity teacher codec to guide the training of StreamCodec2 through knowledge distillation. Experimental results demonstrate that our proposed StreamCodec2, trained with the knowledge distillation strategy, can achieve high-quality speech reconstruction while maintaining low latency (only 20 ms), low computational complexity (only 910 MFLOPs), and low model complexity (only 5.4 M parameters).

eess.AS↗

arXiv2025

Neural Speech Separation with Parallel Amplitude and Phase Spectrum Estimation

This paper proposes APSS, a novel neural speech separation model with parallel amplitude and phase spectrum estimation. Unlike most existing speech separation methods, the APSS distinguishes itself by explicitly estimating the phase spectrum for more complete and accurate separation. Specifically, APSS first extracts the amplitude and phase spectra from the mixed speech signal. Subsequently, the extracted amplitude and phase spectra are fused by a feature combiner into joint representations, which are then further processed by a deep processor with time-frequency Transformers to capture temporal and spectral dependencies. Finally, leveraging parallel amplitude and phase separators, the APSS estimates the respective spectra for each speaker from the resulting features, which are then combined via inverse short-time Fourier transform (iSTFT) to reconstruct the separated speech signals. Experimental results indicate that APSS surpasses both time-domain separation methods and implicit-phase-estimation-based time-frequency approaches. Also, APSS achieves stable and competitive results on multiple datasets, highlighting its strong generalization capability and practical applicability.

cs.SD↗

arXiv2025

Speaker Privacy and Security in the Big Data Era: Protection and Defense against Deepfake

In the era of big data, remarkable advancements have been achieved in personalized speech generation techniques that utilize speaker attributes, including voice and speaking style, to generate deepfake speech. This has also amplified global security risks from deepfake speech misuse, resulting in considerable societal costs worldwide. To address the security threats posed by deepfake speech, techniques have been developed focusing on both the protection of voice attributes and the defense against deepfake speech. Among them, the voice anonymization technique has been developed to protect voice attributes from extraction for deepfake generation, while deepfake detection and watermarking have been utilized to defend against the misuse of deepfake speech. This paper provides a short and concise overview of the three techniques, describing the methodologies, advancements, and challenges. A comprehensive version, offering additional discussions, will be published in the near future.

eess.AS↗

arXiv2025

The First Voice Timbre Attribute Detection Challenge

The first voice timbre attribute detection challenge is featured in a special session at NCMMSC 2025. It focuses on the explainability of voice timbre and compares the intensity of two speech utterances in a specified timbre descriptor dimension. The evaluation was conducted on the VCTK-RVA dataset. Participants developed their systems and submitted their outputs to the organizer, who evaluated the performance and sent feedback to them. Six teams submitted their outputs, with five providing descriptions of their methodologies.

cs.SD↗

arXiv2025

Is GAN Necessary for Mel-Spectrogram-based Neural Vocoder?

Recently, mainstream mel-spectrogram-based neural vocoders rely on generative adversarial network (GAN) for high-fidelity speech generation, e.g., HiFi-GAN and BigVGAN. However, the use of GAN restricts training efficiency and model complexity. Therefore, this paper proposes a novel FreeGAN vocoder, aiming to answer the question of whether GAN is necessary for mel-spectrogram-based neural vocoders. The FreeGAN employs an amplitude-phase serial prediction framework, eliminating the need for GAN training. It incorporates amplitude prior input, SNAKE-ConvNeXt v2 backbone and frequency-weighted anti-wrapping phase loss to compensate for the performance loss caused by the absence of GAN. Experimental results confirm that the speech quality of FreeGAN is comparable to that of advanced GAN-based vocoders, while significantly improving training efficiency and complexity. Other explicit-phase-prediction-based neural vocoders can also work without GAN, leveraging our proposed methods.

eess.AS↗

arXiv2025

The Voice Timbre Attribute Detection 2025 Challenge Evaluation Plan

Voice timbre refers to the unique quality or character of a person's voice that distinguishes it from others as perceived by human hearing. The Voice Timbre Attribute Detection (VtaD) 2025 challenge focuses on explaining the voice timbre attribute in a comparative manner. In this challenge, the human impression of voice timbre is verbalized with a set of sensory descriptors, including bright, coarse, soft, magnetic, and so on. The timbre is explained from the comparison between two voices in their intensity within a specific descriptor dimension. The VtaD 2025 challenge starts in May and culminates in a special proposal at the NCMMSC2025 conference in October 2025 in Zhenjiang, China.

cs.SD↗

arXiv2025

Introducing voice timbre attribute detection

This paper focuses on explaining the timbre conveyed by speech signals and introduces a task termed voice timbre attribute detection (vTAD). In this task, voice timbre is explained with a set of sensory attributes describing its human perception. A pair of speech utterances is processed, and their intensity is compared in a designated timbre descriptor. Moreover, a framework is proposed, which is built upon the speaker embeddings extracted from the speech utterances. The investigation is conducted on the VCTK-RVA dataset. Experimental examinations on the ECAPA-TDNN and FACodec speaker encoders demonstrated that: 1) the ECAPA-TDNN speaker encoder was more capable in the seen scenario, where the testing speakers were included in the training set; 2) the FACodec speaker encoder was superior in the unseen scenario, where the testing speakers were not part of the training, indicating enhanced generalization capability. The VCTK-RVA dataset and open-source code are available on the website https://github.com/vTAD2025-Challenge/vTAD.

cs.SD↗

arXiv2025

ERVQ: Enhanced Residual Vector Quantization with Intra-and-Inter-Codebook Optimization for Neural Audio Codecs

Current neural audio codecs typically use residual vector quantization (RVQ) to discretize speech signals. However, they often experience codebook collapse, which reduces the effective codebook size and leads to suboptimal performance. To address this problem, we introduce ERVQ, Enhanced Residual Vector Quantization, a novel enhancement strategy for the RVQ framework in neural audio codecs. ERVQ mitigates codebook collapse and boosts codec performance through both intra- and inter-codebook optimization. Intra-codebook optimization incorporates an online clustering strategy and a code balancing loss to ensure balanced and efficient codebook utilization. Inter-codebook optimization improves the diversity of quantized features by minimizing the similarity between successive quantizations. Our experiments show that ERVQ significantly enhances audio codec performance across different models, sampling rates, and bitrates, achieving superior quality and generalization capabilities. It also achieves 100% codebook utilization on one of the most advanced neural audio codecs. Further experiments indicate that audio codecs improved by the ERVQ strategy can improve unified speech-and-text large language models (LLMs). Specifically, there is a notable improvement in the naturalness of generated speech in downstream zero-shot text-to-speech tasks. Audio samples are available here.

eess.AS↗

arXiv2025

Universal Preference-Score-based Pairwise Speech Quality Assessment

To compare the performance of two speech generation systems, one of the most effective approaches is estimating the preference score between their generated speech. This paper proposes a novel universal preference-score-based pairwise speech quality assessment (UPPSQA) model, aimed at predicting the preference score between paired speech samples to determine which one has better quality. The model first predicts the absolute mean opinion score (MOS) for the two speech samples separately, and then aggregates them into a relative preference score using a preference function. To address the scarcity of preference data, we also construct a new pairwise speech dataset based on a MOS dataset for experiments. Experimental results confirm that, whether in training scenarios with different data types and label conditions, or in both in-domain and out-of-domain test scenarios, the prediction accuracy of UPP-SQA outperforms that of the baseline models, demonstrating its universality.

cs.SD↗

arXiv2025

Vision-Integrated High-Quality Neural Speech Coding

This paper proposes a novel vision-integrated neural speech codec (VNSC), which aims to enhance speech coding quality by leveraging visual modality information. In VNSC, the image analysis-synthesis module extracts visual features from lip images, while the feature fusion module facilitates interaction between the image analysis-synthesis module and the speech coding module, transmitting visual information to assist the speech coding process. Depending on whether visual information is available during the inference stage, the feature fusion module integrates visual features into the speech coding module using either explicit integration or implicit distillation strategies. Experimental results confirm that integrating visual information effectively improves the quality of the decoded speech and enhances the noise robustness of the neural speech codec, without increasing the bitrate.

eess.AS↗

arXiv2025

Leveraging Cascaded Binary Classification and Multimodal Fusion for Dementia Detection through Spontaneous Speech

This paper presents our submission to the PROCESS Challenge 2025, focusing on spontaneous speech analysis for early dementia detection. For the three-class classification task (Healthy Control, Mild Cognitive Impairment, and Dementia), we propose a cascaded binary classification framework that fine-tunes pre-trained language models and incorporates pause encoding to better capture disfluencies. This design streamlines multi-class classification and addresses class imbalance by restructuring the decision process. For the Mini-Mental State Examination score regression task, we develop an enhanced multimodal fusion system that combines diverse acoustic and linguistic features. Separate regression models are trained on individual feature sets, with ensemble learning applied through score averaging. Experimental results on the test set outperform the baselines provided by the organizers in both tasks, demonstrating the robustness and effectiveness of our approach.

eess.AS↗

arXiv2025

Decoding Speaker-Normalized Pitch from EEG for Mandarin Perception

The same speech content produced by different speakers exhibits significant differences in pitch contour, yet listeners' semantic perception remains unaffected. This phenomenon may stem from the brain's perception of pitch contours being independent of individual speakers' pitch ranges. In this work, we recorded electroencephalogram (EEG) while participants listened to Mandarin monosyllables with varying tones, phonemes, and speakers. The CE-ViViT model is proposed to decode raw or speaker-normalized pitch contours directly from EEG. Experimental results demonstrate that the proposed model can decode pitch contours with modest errors, achieving performance comparable to state-of-the-art EEG regression methods. Moreover, speaker-normalized pitch contours were decoded more accurately, supporting the neural encoding of relative pitch.

cs.SD↗

arXiv2025

Beyond Manual Transcripts: The Potential of Automated Speech Recognition Errors in Improving Alzheimer's Disease Detection

Recent breakthroughs in Automatic Speech Recognition (ASR) have enabled fully automated Alzheimer's Disease (AD) detection using ASR transcripts. Nonetheless, the impact of ASR errors on AD detection remains poorly understood. This paper fills the gap. We conduct a comprehensive study on AD detection using transcripts from various ASR models and their synthesized speech on the ADReSS dataset. Experimental results reveal that certain ASR transcripts (ASR-synthesized speech) outperform manual transcripts (manual-synthesized speech) in detection accuracy, suggesting that ASR errors may provide valuable cues for improving AD detection. Additionally, we propose a cross-attention-based interpretability model that not only identifies these cues but also achieves superior or comparable performance to the baseline. Furthermore, we utilize this model to unveil AD-related patterns within pre-trained embeddings. Our study offers novel insights into the potential of ASR models for AD detection.

eess.AS↗

arXiv2025

Improving Noise Robustness of LLM-based Zero-shot TTS via Discrete Acoustic Token Denoising

Large language model (LLM) based zero-shot text-to-speech (TTS) methods tend to preserve the acoustic environment of the audio prompt, leading to degradation in synthesized speech quality when the audio prompt contains noise. In this paper, we propose a novel neural codec-based speech denoiser and integrate it with the advanced LLM-based TTS model, LauraTTS, to achieve noise-robust zero-shot TTS. The proposed codec denoiser consists of an audio codec, a token denoiser, and an embedding refiner. The token denoiser predicts the first two groups of clean acoustic tokens from the noisy ones, which can serve as the acoustic prompt for LauraTTS to synthesize high-quality personalized speech or be converted to clean speech waveforms through the embedding refiner and codec decoder. Experimental results show that our proposed codec denoiser outperforms state-of-the-art speech enhancement (SE) methods, and the proposed noise-robust LauraTTS surpasses the approach using additional SE models.

eess.AS↗

arXiv2025

Weakly Supervised Temporal Sentence Grounding via Positive Sample Mining

The task of weakly supervised temporal sentence grounding (WSTSG) aims to detect temporal intervals corresponding to a language description from untrimmed videos with only video-level video-language correspondence. For an anchor sample, most existing approaches generate negative samples either from other videos or within the same video for contrastive learning. However, some training samples are highly similar to the anchor sample, directly regarding them as negative samples leads to difficulties for optimization and ignores the correlations between these similar samples and the anchor sample. To address this, we propose Positive Sample Mining (PSM), a novel framework that mines positive samples from the training set to provide more discriminative supervision. Specifically, for a given anchor sample, we partition the remaining training set into semantically similar and dissimilar subsets based on the similarity of their text queries. To effectively leverage these correlations, we introduce a PSM-guided contrastive loss to ensure that the anchor proposal is closer to similar samples and further from dissimilar ones. Additionally, we design a PSM-guided rank loss to ensure that similar samples are closer to the anchor proposal than to the negative intra-video proposal, aiming to distinguish the anchor proposal and the negative intra-video proposal. Experiments on the WSTSG and grounded VideoQA tasks demonstrate the effectiveness and superiority of our method.

cs.CV↗

arXiv2025

A Streamable Neural Audio Codec with Residual Scalar-Vector Quantization for Real-Time Communication

This paper proposes StreamCodec, a streamable neural audio codec designed for real-time communication. StreamCodec adopts a fully causal, symmetric encoder-decoder structure and operates in the modified discrete cosine transform (MDCT) domain, aiming for low-latency inference and real-time efficient generation. To improve codebook utilization efficiency and compensate for the audio quality loss caused by structural causality, StreamCodec introduces a novel residual scalar-vector quantizer (RSVQ). The RSVQ sequentially connects scalar quantizers and improved vector quantizers in a residual manner, constructing coarse audio contours and refining acoustic details, respectively. Experimental results confirm that the proposed StreamCodec achieves decoded audio quality comparable to advanced non-streamable neural audio codecs. Specifically, on the 16 kHz LibriTTS dataset, StreamCodec attains a ViSQOL score of 4.30 at 1.5 kbps. It has a fixed latency of only 20 ms and achieves a generation speed nearly 20 times real-time on a CPU, with a lightweight model size of just 7M parameters, making it highly suitable for real-time communication applications.

cs.SD↗

arXiv2025

Perturbation-Restrained Sequential Model Editing

Model editing is an emerging field that focuses on updating the knowledge embedded within large language models (LLMs) without extensive retraining. However, current model editing methods significantly compromise the general abilities of LLMs as the number of edits increases, and this trade-off poses a substantial challenge to the continual learning of LLMs. In this paper, we first theoretically analyze that the factor affecting the general abilities in sequential model editing lies in the condition number of the edited matrix. The condition number of a matrix represents its numerical sensitivity, and therefore can be used to indicate the extent to which the original knowledge associations stored in LLMs are perturbed after editing. Subsequently, statistical findings demonstrate that the value of this factor becomes larger as the number of edits increases, thereby exacerbating the deterioration of general abilities. To this end, a framework termed Perturbation Restraint on Upper bouNd for Editing (PRUNE) is proposed, which applies the condition number restraints in sequential editing. These restraints can lower the upper bound on perturbation to edited models, thus preserving the general abilities. Systematically, we conduct experiments employing three editing methods on three LLMs across four downstream tasks. The results show that PRUNE can preserve general abilities while maintaining the editing performance effectively in sequential model editing. The code are available at https://github.com/mjy1111/PRUNE.

cs.CL↗