arXiv subjects

Lei Xie

Publications and source records attributed to Lei Xie.

At least 127 records · Page 7Linked to original sources

arXiv2025

Fine-grained Preference Optimization Improves Zero-shot Text-to-Speech

Integrating human feedback to align text-to-speech (TTS) system outputs with human preferences has proven to be an effective approach for enhancing the robustness of language model-based TTS systems. Current approaches primarily focus on using preference data annotated at the utterance level. However, frequent issues that affect the listening experience often only arise in specific segments of audio samples, while other segments are well-generated. In this study, we propose a fine-grained preference optimization approach (FPO) to enhance the robustness of TTS systems. FPO focuses on addressing localized issues in generated samples rather than uniformly optimizing the entire utterance. Specifically, we first analyze the types of issues in generated samples, categorize them into two groups, and propose a selective training loss strategy to optimize preferences based on fine-grained labels for each issue type. Experimental results show that FPO enhances the robustness of zero-shot TTS systems by effectively addressing local issues, significantly reducing the bad case ratio, and improving intelligibility. Furthermore, FPO exhibits superior data efficiency compared with baseline systems, achieving similar performance with fewer training samples.

eess.AS↗

arXiv2025

MeanVC: Lightweight and Streaming Zero-Shot Voice Conversion via Mean Flows

Zero-shot voice conversion (VC) aims to transfer timbre from a source speaker to any unseen target speaker while preserving linguistic content. Growing application scenarios demand models with streaming inference capabilities. This has created a pressing need for models that are simultaneously fast, lightweight, and high-fidelity. However, existing streaming methods typically rely on either autoregressive (AR) or non-autoregressive (NAR) frameworks, which either require large parameter sizes to achieve strong performance or struggle to generalize to unseen speakers. In this study, we propose MeanVC, a lightweight and streaming zero-shot VC approach. MeanVC introduces a diffusion transformer with a chunk-wise autoregressive denoising strategy, combining the strengths of both AR and NAR paradigms for efficient streaming processing. By introducing mean flows, MeanVC regresses the average velocity field during training, enabling zero-shot VC with superior speech quality and speaker similarity in a single sampling step by directly mapping from the start to the endpoint of the flow trajectory. Additionally, we incorporate diffusion adversarial post-training to mitigate over-smoothing and further enhance speech quality. Experimental results demonstrate that MeanVC significantly outperforms existing zero-shot streaming VC systems, achieving superior conversion quality with higher efficiency and significantly fewer parameters. Audio demos and code are publicly available at https://aslp-lab.github.io/MeanVC.

eess.AS↗

arXiv2025

X-Talk: On the Underestimated Potential of Modular Speech-to-Speech Dialogue System

We present X-Talk, an open-source framework that champions a decoupled, modular design for LLM-driven speech-to-speech (S2S) systems. While the dominant trend favors end-to-end (E2E) modeling to optimize information flow, these "omni-models" often struggle to balance the competing objectives of complex speech tasks within a single network. X-Talk challenges this paradigm by demonstrating that a systematically optimized cascaded pipeline can achieve sub-second latency without sacrificing modular flexibility. Our framework seamlessly integrates specialized front-end components (e.g., VAD, speech enhancement) and diverse understanding models (e.g., ASR, emotion, and environmental sound analysis) with LLM capabilities like retrieval-augmented generation (RAG) and tool use. By revitalizing the cascaded approach, X-Talk highlights the underestimated potential of modular S2S systems and provides a robust foundation for future research and applications.

cs.SD↗

arXiv2025

Adaptive Matched Filtering for Sensing With Communication Signals in Cluttered Environments

This paper investigates the performance of the adaptive matched filtering (AMF) in cluttered environments, particularly when operating with superimposed signals. Since the instantaneous signal-to-clutter-plus-noise ratio (SCNR) is a random variable dependent on the data payload, using it directly as a design objective poses severe practical challenges, such as prohibitive computational burdens and signaling overhead. To address this, we propose shifting the optimization objective from an instantaneous to a statistical metric, which focuses on maximizing the average SCNR over all possible payloads. Due to its analytical intractability, we leverage tools from random matrix theory (RMT) to derive an asymptotic approximation for the average SCNR, which remains accurate even in moderate-dimensional regimes. A key finding from our theoretical analysis is that, for a fixed modulation basis, the PSK achieves a superior average SCNR compared to QAM and the pure Gaussian constellation. Furthermore, for any given constellation, the OFDM achieves a higher average SCNR than SC and AFDM. Then, we propose two pilot design schemes to enhance system performance: a Data-Payload-Dependent (DPD) scheme and a Data-Payload-Independent (DPI) scheme. The DPD approach maximizes the instantaneous SCNR for each transmission. Conversely, the DPI scheme optimizes the average SCNR, offering a flexible trade-off between sensing performance and implementation complexity. Then, we develop two dedicated optimization algorithms for DPD and DPI schemes. In particular, for the DPD problem, we employ fractional optimization and the KKT conditions to derive a closed-form solution. For the DPI problem, we adopt a manifold optimization approach to handle the inherent rank-one constraint efficiently. Simulation results validate the accuracy of our theoretical analysis and demonstrate the effectiveness of the proposed methods.

cs.IT↗

arXiv2025

YingMusic-Singer: Zero-shot Singing Voice Synthesis and Editing with Annotation-free Melody Guidance

Singing Voice Synthesis (SVS) remains constrained in practical deployment due to its strong dependence on accurate phoneme-level alignment and manually annotated melody contours, requirements that are resource-intensive and hinder scalability. To overcome these limitations, we propose a melody-driven SVS framework capable of synthesizing arbitrary lyrics following any reference melody, without relying on phoneme-level alignment. Our method builds on a Diffusion Transformer (DiT) architecture, enhanced with a dedicated melody extraction module that derives melody representations directly from reference audio. To ensure robust melody encoding, we employ a teacher model to guide the optimization of the melody extractor, alongside an implicit alignment mechanism that enforces similarity distribution constraints for improved melodic stability and coherence. Additionally, we refine duration modeling using weakly annotated song data and introduce a Flow-GRPO reinforcement learning strategy with a multi-objective reward function to jointly enhance pronunciation clarity and melodic fidelity. Experiments show that our model achieves superior performance over existing approaches in both objective measures and subjective listening tests, especially in zero-shot and lyric adaptation settings, while maintaining high audio quality without manual annotation. This work offers a practical and scalable solution for advancing data-efficient singing voice synthesis. To support reproducibility, we release our inference code and model checkpoints.

cs.SD↗

arXiv2025

WEST: LLM based Speech Toolkit for Speech Understanding, Generation, and Interaction

In this paper, we present WEST(WE Speech Toolkit), a speech toolkit based on a large language model (LLM) for speech understanding, generation, and interaction. There are three key features of WEST: 1) Fully LLM-based: Standing on the shoulders of giants by reusing mature architectures, ecosystems (e.g., Hugging Face), and methods (e.g., sequence packing) from large models. 2) Full-stack: Supports tasks such as recognition, synthesis, understanding, dialogue, and multimodal capabilities, with extensibility to incorporate open-source models. 3) Simple and Stupid: A simple and stupid speech toolkit that everyone can Touch. In addition, WEST provides two types of recipes, models, and experimental results. The first is entirely based on open-source models and open-source data, allowing users to fully reproduce the experiments in this paper and serving as a verification system or minimal system baseline. The second is trained on massive data, offering superior performance so the user can directly apply it out of the box. WEST is publicly avilable at https://github.com/wenet-e2e/west/

cs.CL↗

arXiv2025

SoulX-Podcast: Towards Realistic Long-form Podcasts with Dialectal and Paralinguistic Diversity

Recent advances in text-to-speech (TTS) synthesis have significantly improved speech expressiveness and naturalness. However, most existing systems are tailored for single-speaker synthesis and fall short in generating coherent multi-speaker conversational speech. This technical report presents SoulX-Podcast, a system designed for podcast-style multi-turn, multi-speaker dialogic speech generation, while also achieving state-of-the-art performance in conventional TTS tasks. To meet the higher naturalness demands of multi-turn spoken dialogue, SoulX-Podcast integrates a range of paralinguistic controls and supports both Mandarin and English, as well as several Chinese dialects, including Sichuanese, Henanese, and Cantonese, enabling more personalized podcast-style speech generation. Experimental results demonstrate that SoulX-Podcast can continuously produce over 90 minutes of conversation with stable speaker timbre and smooth speaker transitions. Moreover, speakers exhibit contextually adaptive prosody, reflecting natural rhythm and intonation changes as dialogues progress. Across multiple evaluation metrics, SoulX-Podcast achieves state-of-the-art performance in both monologue TTS and multi-turn conversational speech synthesis.

eess.AS↗

arXiv2025

Adaptive Individual Uncertainty under Out-Of-Distribution Shift with Expert-Routed Conformal Prediction

Reliable, informative, and individual uncertainty quantification (UQ) remains missing in current ML community. This hinders the effective application of AI/ML to risk-sensitive domains. Most methods either fail to provide coverage on new data, inflate intervals so broadly that they are not actionable, or assign uncertainties that do not track actual error, especially under a distribution shift. In high-stakes drug discovery, protein-ligand affinity (PLI) prediction is especially challenging as assay noise is heterogeneous, chemical space is imbalanced and large, and practical evaluations routinely involve distribution shift. In this work, we introduce a novel uncertainty quantification method, Trustworthy Expert Split-conformal with Scaled Estimation for Efficient Reliable Adaptive intervals (TESSERA), that provides per-sample uncertainty with reliable coverage guarantee, informative and adaptive prediction interval widths that track the absolute error. We evaluate on protein-ligand binding affinity prediction under both independent and identically distributed (i.i.d.) and scaffold-based out-of-distribution (OOD) splits, comparing against strong UQ baselines. TESSERA attains near-nominal coverage and the best coverage-width trade-off as measured by the Coverage-Width Criterion (CWC), while maintaining competitive adaptivity (lowest Area Under the Sparsification Error (AUSE)). Size-Stratified Coverage (SSC) further confirms that intervals are right-sized, indicating width increases when data are scarce or noisy, and remain tight when predictions are reliable. By unifying Mixture of Expert (MoE) diversity with conformal calibration, TESSERA delivers trustworthy, tight, and adaptive uncertainties that are well-suited to selective prediction and downstream decision-making in the drug-discovery pipeline and other applications.

cs.LG↗

arXiv2025

LSZone: A Lightweight Spatial Information Modeling Architecture for Real-time In-car Multi-zone Speech Separation

In-car multi-zone speech separation, which captures voices from different speech zones, plays a crucial role in human-vehicle interaction. Although previous SpatialNet has achieved notable results, its high computational cost still hinders real-time applications in vehicles. To this end, this paper proposes LSZone, a lightweight spatial information modeling architecture for real-time in-car multi-zone speech separation. We design a spatial information extraction-compression (SpaIEC) module that combines Mel spectrogram and Interaural Phase Difference (IPD) to reduce computational burden while maintaining performance. Additionally, to efficiently model spatial information, we introduce an extremely lightweight Conv-GRU crossband-narrowband processing (CNP) module. Experimental results demonstrate that LSZone, with a complexity of 0.56G MACs and a real-time factor (RTF) of 0.37, delivers impressive performance in complex noise and multi-speaker scenarios.

cs.SD↗

arXiv2025

SynthVC: Leveraging Synthetic Data for End-to-End Low Latency Streaming Voice Conversion

Voice Conversion (VC) aims to modify a speaker's timbre while preserving linguistic content. While recent VC models achieve strong performance, most struggle in real-time streaming scenarios due to high latency, dependence on ASR modules, or complex speaker disentanglement, which often results in timbre leakage or degraded naturalness. We present SynthVC, a streaming end-to-end VC framework that directly learns speaker timbre transformation from synthetic parallel data generated by a pre-trained zero-shot VC model. This design eliminates the need for explicit content-speaker separation or recognition modules. Built upon a neural audio codec architecture, SynthVC supports low-latency streaming inference with high output fidelity. Experimental results show that SynthVC outperforms baseline streaming VC systems in both naturalness and speaker similarity, achieving an end-to-end latency of just 77.1 ms.

cs.SD↗

arXiv2025

Serial-Parallel Dual-Path Architecture for Speaking Style Recognition

Speaking Style Recognition (SSR) identifies a speaker's speaking style characteristics from speech. Existing style recognition approaches primarily rely on linguistic information, with limited integration of acoustic information, which restricts recognition accuracy improvements. The fusion of acoustic and linguistic modalities offers significant potential to enhance recognition performance. In this paper, we propose a novel serial-parallel dual-path architecture for SSR that leverages acoustic-linguistic bimodal information. The serial path follows the ASR+STYLE serial paradigm, reflecting a sequential temporal dependency, while the parallel path integrates our designed Acoustic-Linguistic Similarity Module (ALSM) to facilitate cross-modal interaction with temporal simultaneity. Compared to the existing SSR baseline -- the OSUM model, our approach reduces parameter size by 88.4% and achieves a 30.3% improvement in SSR accuracy for eight styles on the test set.

cs.SD↗

arXiv2025

DialoSpeech: Dual-Speaker Dialogue Generation with LLM and Flow Matching

Recent advances in text-to-speech (TTS) synthesis, particularly those leveraging large language models (LLMs), have significantly improved expressiveness and naturalness. However, generating human-like, interactive dialogue speech remains challenging. Current systems face limitations due to the scarcity of dual-track data and difficulties in achieving naturalness, contextual coherence, and interactional dynamics, such as turn-taking, overlapping speech, and speaker consistency, in multi-turn conversations. To address these challenges, we propose DialoSpeech, a dual-track architecture combining a large language model with Chunked Flow Matching for expressive, human-like dialogue speech synthesis. DialoSpeech generates natural multi-turn conversations with coherent speaker turns and natural overlaps, supporting both Chinese and English and cross-lingual speech synthesis. We introduce a data processing pipeline to construct dual-track dialogue datasets, facilitating scalable training and experimental validation. Experiments show that our model outperforms baselines, offering a solution for generating human-like spoken dialogues. Audio samples are available at https://tiamojames.github.io/DialoSpeech

eess.AS↗

arXiv2025

MeanFlowSE: One-Step Generative Speech Enhancement via MeanFlow

Speech enhancement (SE) recovers clean speech from noisy signals and is vital for applications such as telecommunications and automatic speech recognition (ASR). While generative approaches achieve strong perceptual quality, they often rely on multi-step sampling (diffusion/flow-matching) or large language models, limiting real-time deployment. To mitigate these constraints, we present MeanFlowSE, a one-step generative SE framework. It adopts MeanFlow to predict an average-velocity field for one-step latent refinement and conditions the model on self-supervised learning (SSL) representations rather than VAE latents. This design accelerates inference and provides robust acoustic-semantic guidance during training. In the Interspeech 2020 DNS Challenge blind test set and simulated test set, MeanFlowSE attains state-of-the-art (SOTA) level perceptual quality and competitive intelligibility while significantly lowering both real-time factor (RTF) and model size compared with recent generative competitors, making it suitable for practical use. The code will be released upon publication at https://github.com/Hello3orld/MeanFlowSE.

cs.SD↗

arXiv2025

HiStyle: Hierarchical Style Embedding Predictor for Text-Prompt-Guided Controllable Speech Synthesis

Controllable speech synthesis refers to the precise control of speaking style by manipulating specific prosodic and paralinguistic attributes, such as gender, volume, speech rate, pitch, and pitch fluctuation. With the integration of advanced generative models, particularly large language models (LLMs) and diffusion models, controllable text-to-speech (TTS) systems have increasingly transitioned from label-based control to natural language description-based control, which is typically implemented by predicting global style embeddings from textual prompts. However, this straightforward prediction overlooks the underlying distribution of the style embeddings, which may hinder the full potential of controllable TTS systems. In this study, we use t-SNE analysis to visualize and analyze the global style embedding distribution of various mainstream TTS systems, revealing a clear hierarchical clustering pattern: embeddings first cluster by timbre and subsequently subdivide into finer clusters based on style attributes. Based on this observation, we propose HiStyle, a two-stage style embedding predictor that hierarchically predicts style embeddings conditioned on textual prompts, and further incorporate contrastive learning to help align the text and audio embedding spaces. Additionally, we propose a style annotation strategy that leverages the complementary strengths of statistical methodologies and human auditory preferences to generate more accurate and perceptually consistent textual prompts for style control. Comprehensive experiments demonstrate that when applied to the base TTS model, HiStyle achieves significantly better style controllability than alternative style embedding predicting approaches while preserving high speech quality in terms of naturalness and intelligibility. Audio samples are available at https://anonymous.4open.science/w/HiStyle-2517/.

cs.AI↗

arXiv2025

Easy Turn: Integrating Acoustic and Linguistic Modalities for Robust Turn-Taking in Full-Duplex Spoken Dialogue Systems

Full-duplex interaction is crucial for natural human-machine communication, yet remains challenging as it requires robust turn-taking detection to decide when the system should speak, listen, or remain silent. Existing solutions either rely on dedicated turn-taking models, most of which are not open-sourced. The few available ones are limited by their large parameter size or by supporting only a single modality, such as acoustic or linguistic. Alternatively, some approaches finetune LLM backbones to enable full-duplex capability, but this requires large amounts of full-duplex data, which remain scarce in open-source form. To address these issues, we propose Easy Turn, an open-source, modular turn-taking detection model that integrates acoustic and linguistic bimodal information to predict four dialogue turn states: complete, incomplete, backchannel, and wait, accompanied by the release of Easy Turn trainset, a 1,145-hour speech dataset designed for training turn-taking detection models. Compared to existing open-source models like TEN Turn Detection and Smart Turn V2, our model achieves state-of-the-art turn-taking detection accuracy on our open-source Easy Turn testset. The data and model will be made publicly available on GitHub.

cs.CL↗

arXiv2025

WenetSpeech-Chuan: A Large-Scale Sichuanese Corpus with Rich Annotation for Dialectal Speech Processing

The scarcity of large-scale, open-source data for dialects severely hinders progress in speech technology, a challenge particularly acute for the widely spoken Sichuanese dialects of Chinese. To address this critical gap, we introduce WenetSpeech-Chuan, a 10,000-hour, richly annotated corpus constructed using our novel Chuan-Pipeline, a complete data processing framework for dialectal speech. To facilitate rigorous evaluation and demonstrate the corpus's effectiveness, we also release high-quality ASR and TTS benchmarks, WenetSpeech-Chuan-Eval, with manually verified transcriptions. Experiments show that models trained on WenetSpeech-Chuan achieve state-of-the-art performance among open-source systems and demonstrate results comparable to commercial services. As the largest open-source corpus for Sichuanese dialects, WenetSpeech-Chuan not only lowers the barrier to research in dialectal speech processing but also plays a crucial role in promoting AI equity and mitigating bias in speech technologies. The corpus, benchmarks, models, and receipts are publicly available on our project page.

cs.CL↗

arXiv2025

Towards Building Speech Large Language Models for Multitask Understanding in Low-Resource Languages

Speech large language models (SLLMs) built on speech encoders, adapters, and LLMs demonstrate remarkable multitask understanding performance in high-resource languages such as English and Chinese. However, their effectiveness substantially degrades in low-resource languages such as Thai. This limitation arises from three factors: (1) existing commonly used speech encoders, like the Whisper family, underperform in low-resource languages and lack support for broader spoken language understanding tasks; (2) the ASR-based alignment paradigm requires training the entire SLLM, leading to high computational cost; (3) paired speech-text data in low-resource languages is scarce. To overcome these challenges in the low-resource language Thai, we introduce XLSR-Thai, the first self-supervised learning (SSL) speech encoder for Thai. It is obtained by continuously training the standard SSL XLSR model on 36,000 hours of Thai speech data. Furthermore, we propose U-Align, a speech-text alignment method that is more resource-efficient and multitask-effective than typical ASR-based alignment. Finally, we present Thai-SUP, a pipeline for generating Thai spoken language understanding data from high-resource languages, yielding the first Thai spoken language understanding dataset of over 1,000 hours. Multiple experiments demonstrate the effectiveness of our methods in building a Thai multitask-understanding SLLM. We open-source XLSR-Thai and Thai-SUP to facilitate future research.

cs.SD↗

arXiv2025

KALL-E:Autoregressive Speech Synthesis with Next-Distribution Prediction

We introduce KALL-E, a novel autoregressive (AR) language model for text-to-speech (TTS) synthesis that operates by predicting the next distribution of continuous speech frames. Unlike existing methods, KALL-E directly models the continuous speech distribution conditioned on text, eliminating the need for any diffusion-based components. Specifically, we utilize a Flow-VAE to extract a continuous latent speech representation from waveforms, instead of relying on discrete speech tokens. A single AR Transformer is then trained to predict these continuous speech distributions from text, optimizing a Kullback-Leibler divergence loss as its objective. Experimental results demonstrate that KALL-E achieves superior speech synthesis quality and can even adapt to a target speaker from just a single sample. Importantly, KALL-E provides a more direct and effective approach for utilizing continuous speech representations in TTS.

eess.AS↗