[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82498-en":3,"doc-seo-82498-105":29,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},82498,1099513958762,"Logic","https://ap-avatar.wpscdn.com/avatar/1000023916a998db790?x-image-process=image/resize,m_fixed,w_180,h_180&k=1782109480056885918",8,"Research & Report","Enhancing Flow Matching with a Unified Guidance Framework for Efficient and Robust Speech Synthesis","Flow Matching (FM) for speech generation is limited by high inference latency and timbre leakage. A unified guidance framework is presented to improve both efficiency and robustness via two complementary mechanisms. Data-guidance uses heterogeneous augmentation to encourage disentanglement between linguistic content and acoustic residue. Enhanced model-guidance combines trajectory rectification with an intrinsic guidance objective, distilling conditional knowledge into network weights and removing Classifier-Free Guidance (CFG) overhead. Experiments show nearly threefold faster inference and better speaker similarity than strong baselines.","Enhancing Flow Matching with A Unified Guidance Framework for Efficient  \nand Robust Speech Synthesis  \nZuda Yu∗, Qianhui Xu∗, Ting Chen∗∗, Junhui Zhang∗∗, Tao Fu∗∗, Hongjiang Yu∗∗, Qiangqing Wang∗∗,  \nYang Song∗∗  \nZuoyebang, China  \n{yuzuda, xuqianhui02, chenting08, zhangjunhui04, futao01, yuhongjiang, wangqiangqiang,  \n[songyang](songyang}@zuoyebang.com)[}](songyang}@zuoyebang.com)[@zuoyebang.com](songyang}@zuoyebang.com)  \narXiv :2607 .00363v 1 [ cs . SD] 1 Jul 2026  \nAbstract  \nFlow Matching (FM) has emerged as a powerful paradigm for speech generation but remains constrained by high inference latency and timbre leakage. To address these bottlenecks, we propose a unified guidance framework that enhances generation efficiency and robustness through two complementary strategies. On the data front, we introduce Data-guidance via heterogeneous augmentation, encouraging the model to disentangle linguistic content from acoustic residue. In parallel, we propose an enhanced Model-guidance mechanism that synergizes trajectory rectification with a novel intrinsic guidance objective. This approach distills conditional knowledge into network weights and straightens inference trajectory path, thereby eliminating Classifier-Free Guidance (CFG) overhead. Experiments demonstrate that our framework accelerates inference by nearly three times while effectively improving speaker similarity compared to state-of-the-art baselines. Audio samples are available here.1.  \nIndex Terms: speech synthesis, voice conversion, flow matching  \n1. Introduction  \nFlow Matching (FM) [1] has rapidly advanced speech synthesis by modeling the continuous transformation from a simple prior to complex data distributions. This paradigm has demonstrated remarkable potential across various speech generation tasks. For instance, text-based approaches such as F5-TTS [2] and Matcha-TTS [3] achieve fully non-autoregressive speech generation by directly mapping text to acoustic features. Concurrently, discrete-token-based frameworks including the CosyVoice family [4–7], MaskGCT [8], and FireRedTTS [9] apply FM to convert semantic representations into mel-spectrograms. Furthermore, recent audio-language models like Kimi-Audio [10], GLM-4-Voice [11], and Step-Audio [12] increasingly rely on FM as a high-fidelity detokenizer for waveform reconstruction. Beyond speech synthesis, models such as Seed-VC [13] and StableVC [14] illustrate the capacity of FM to effectively disentangle speech attributes for voice conversion. Despite these impressive milestones, the widespread deployment of FM models in real-time scenarios remains fundamentally constrained by two distinct yet critical bottlenecks.  \nThe first bottleneck, timbre leakage, compromises the robustness of zero-shot generation. Token-based models rely on discrete semantic representations as conditioning inputs, yet  \n*These authors contributed equally.  \n**indicates the corresponding author.  \n1 [https://yuzuda283.github.io/unified-guidanc](https://yuzuda283.github.io/unified-guidanc)e-flow-matching/Interspeech2026_demo_samples/  \nextracting purely disentangled content is notoriously difficult. Traditional information bottlenecks such as Vector Quantization (VQ) [15] attempt to strip away speaker identity but often degrade linguistic intelligibility and pronunciation quality. Consequently, data-driven approaches have gained prominence. SeedTTS [16] mitigates leakage by retraining models on synthetic cross-speaker pairs generated via self-distillation, while SeedVC [13] successfully employs an external generative model to perturb the source audio during training, forcing the network to align with the target prompt. These strategies inspire us to extend single-stage perturbation into a more comprehensive dataguidance constraint.  \nThe second bottleneck, inference latency, stems from the intrinsic nature of ODE-based generation imposing two distinct computational costs. First, iterative ODE solvers require a high Number of F","cbCaipMGLxzxcYvs","https://ap.wps.com/l/cbCaipMGLxzxcYvs","pdf",380380,1,6,"English","en",105,"# Abstract\n# Introduction\n## Timbre leakage in zero-shot generation\n## Inference latency in ODE-based generation\n## Unified guidance framework overview","[{\"question\":\"What problem does the unified guidance framework address in flow matching speech synthesis?\",\"answer\":\"It targets two bottlenecks: high inference latency and timbre leakage, which reduce robustness in zero-shot generation.\"},{\"question\":\"How does Data-guidance reduce timbre leakage?\",\"answer\":\"It uses dual-stage heterogeneous augmentation with model-driven cross-synthesis and signal-driven acoustic deformations to weaken the acoustic reliability of semantic tokens during training.\"},{\"question\":\"How does Enhanced Model-guidance improve efficiency and avoid CFG overhead?\",\"answer\":\"It synergizes trajectory rectification with an intrinsic guidance objective, distilling conditional knowledge into network weights to straighten the inference trajectory and eliminate Classifier-Free Guidance’s extra conditional/unconditional passes.\"}]",1784180934,15,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"enhancing-flow-matching-with-a-unified-guidance-framework-for-efficient-and-robust-speech-synthesis","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/enhancing-flow-matching-with-a-unified-guidance-framework-for-efficient-and-robust-speech-synthesis/82498/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does the unified guidance framework address in flow matching speech synthesis?","Question",{"text":75,"@type":76},"It targets two bottlenecks: high inference latency and timbre leakage, which reduce robustness in zero-shot generation.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does Data-guidance reduce timbre leakage?",{"text":80,"@type":76},"It uses dual-stage heterogeneous augmentation with model-driven cross-synthesis and signal-driven acoustic deformations to weaken the acoustic reliability of semantic tokens during training.",{"name":82,"@type":73,"acceptedAnswer":83},"How does Enhanced Model-guidance improve efficiency and avoid CFG overhead?",{"text":84,"@type":76},"It synergizes trajectory rectification with an intrinsic guidance objective, distilling conditional knowledge into network weights to straighten the inference trajectory and eliminate Classifier-Free Guidance’s extra conditional/unconditional passes.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":21,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]