[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84455-en":3,"doc-seo-84455-105":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},84455,1099513958762,"Logic","https://ap-avatar.wpscdn.com/avatar/1000023916a998db790?x-image-process=image/resize,m_fixed,w_180,h_180&k=1782109480056885918",8,"Research & Report","MoLingo: Motion-Language Alignment for Text-to-Human Motion Generation","MoLingo presents a text-to-motion model that generates realistic, lifelike human movement using denoising in a continuous latent space. The work examines how to make diffusion on motion latents effective, addressing both semantically aligned latent representations and the injection of text conditioning. A semantic-aligned motion encoder trained with frame-level text labels keeps latents with similar meanings close, improving diffusion suitability. Multi-token cross-attention yields stronger text–motion alignment and motion realism, establishing state-of-the-art results on standard metrics and in a user study.","arXiv :2512 . 13840v3 [ cs .CV] 13 Jul 2026  \nMoLingo: Motion–Language Alignment for Text-to-Human Motion Generation  \nYannan He 1,2 Garvita Tiwari 1,2,3 Xiaohan Zhang 1,2,3 Pankaj Bora 1,5  \nTolga Birdal 4 Jan Eric Lenssen 3 Gerard Pons-Moll 1,2,3  \n1University of Tübingen, Germany, 2 Tübingen AI Center, Germany  \n3Max Planck Institute for Informatics, Saarland Informatics Campus, Germany  \n4Imperial College London, UK, 5Zuse School ELIZA, Germany  \n[https://hynann.github.io/molingo/MoLingo.html](https://hynann.github.io/molingo/MoLingo.html)  \nA person swings a golf club.  \nA person does a box step The person is dribbling a  \nwaltz [backwards.](backwards. basketball)[ basketball](backwards. basketball) backwards  \nA figure dances ballet elegantly. The person is doing a dance twirl. The person is sweeping the floor.  \nFigure 1 . Left: Given text prompts, MoLingo generates realistic and text-aligned motions, ranging from daily movements like sweeping to more challenging movements like dancing. Right: MoLingo significantly outperforms previous works in both FID and R-Precision scores. The difference can best be seen in motion, hence we urge the reader to view our project page.  \nAbstract  \nWe introduce MoLingo, a text-to-motion (T2M) model that generates realistic, lifelike human motion by denoising in a continuous latent space. Recent works perform latent space diffusion, either on the whole latent at once or auto-regressively over multiple latents. In this paper, we study how to make diffusion on continuous motion latents work best. We focus on two questions: (1) how to build a semantically aligned latent space so diffusion becomes more effective, and (2) how to best inject text conditioning so the motion follows the description closely. We propose a semantic-aligned motion encoder trained with frame-level text labels so that latents with similar text meaning stay close, which makes the latent space more diffusion-friendly. We also compare single-token conditioning with a multi-token cross-attention scheme and find that cross-attention gives better motion realism and text–motion alignment. With semantically aligned latents, auto-regressive generation, and cross-attention text conditioning, our model sets a new state-  \nof-the-art in human motion generation on standard metricsand in a user study. We will release our code and models for further research and downstream usage.  \n1. Introduction  \nGenerating realistic human motion from text is crucial for computer animation, AR/VR entertainment, and building agents that can follow human instructions. In recent years, text-labeled human motion datasets [16, 51] have kicked off this research area, and diffusion-based generative models have brought a major improvement in generated motion quality. Earlier text-to-motion (T2M) works used diffusion models to denoise motion directly in pose space [4, 8, 13, 23, 27, 31, 35, 58, 75, 76, 87] and achieved good performance. However, denoising raw pose frames is hard because the joint distribution is very complex, and it can also brings artifacts by preserving mocap noise [41] . To address this, recent  \nworks first encode motion into a compact latent space [6, 9, 10, 84, 85], then run diffusion in that space, and finally decode it back to motion. Current methods mainly follow two tracks: (1) diffusion on the whole latent space at once [6, 9, 10, 26], and (2) split the sequence into multiple latentsand denoise them auto-regressively [43, 61, 69, 86, 88] .  \nIn this work, we investigate how to most effectively perform text-to-motion diffusion in a continuous latent space. There are two key questions here: (1) what makes a good latent space for motion diffusion, and (2) how to inject the text condition most effectively.  \nFor the first question, we study motion latent spaces w.r.t. latent size and semantic alignment, motivated by recent latent space studies for image generation [29, 67, 70] . Specifically, we introduce a semantically aligned mot","cbCaikT2utd9Yrzi","https://ap.wps.com/l/cbCaikT2utd9Yrzi","pdf",3608310,1,17,"English","en",105,"# Introduction\n# Related Work","[{\"question\":\"What problem does MoLingo address in text-to-human motion generation?\",\"answer\":\"MoLingo tackles how to perform diffusion effectively in a continuous motion latent space while ensuring the generated motion stays closely aligned with the input text description.\"},{\"question\":\"How does MoLingo improve the motion latent space for diffusion?\",\"answer\":\"It proposes a semantic-aligned motion encoder trained with frame-level text labels so motion latents with similar textual meanings are pulled closer, making diffusion more effective and instruction-following more faithful.\"},{\"question\":\"Why does MoLingo use multi-token cross-attention for text conditioning instead of a single token?\",\"answer\":\"Experiments show that a single text token lacks expressiveness; multi-token conditioning with cross-attention provides stronger conditioning, improving both motion realism and text–motion alignment.\"}]",1784195727,43,{"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},"molingo-motion-language-alignment-for-text-to-human-motion-generation","",{"@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/molingo-motion-language-alignment-for-text-to-human-motion-generation/84455/",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 MoLingo address in text-to-human motion generation?","Question",{"text":75,"@type":76},"MoLingo tackles how to perform diffusion effectively in a continuous motion latent space while ensuring the generated motion stays closely aligned with the input text description.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does MoLingo improve the motion latent space for diffusion?",{"text":80,"@type":76},"It proposes a semantic-aligned motion encoder trained with frame-level text labels so motion latents with similar textual meanings are pulled closer, making diffusion more effective and instruction-following more faithful.",{"name":82,"@type":73,"acceptedAnswer":83},"Why does MoLingo use multi-token cross-attention for text conditioning instead of a single token?",{"text":84,"@type":76},"Experiments show that a single text token lacks expressiveness; multi-token conditioning with cross-attention provides stronger conditioning, improving both motion realism and text–motion alignment.","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},[93,97,101,105,110,115,120,123,128,131,135],{"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":111,"doc_module":4,"doc_module_name":45,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":45,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":45,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]