Dong yu tencent In this paper, we propose novel bilateral denoising diffusion models (BDDMs), which take signif- In this paper, we propose a Chinese multi-turn topic-driven conversation dataset, NaturalConv, which allows the participants to chat anything they want as long as any element from the topic is mentioned and the topic shift is smooth. 2024. alpha@gmail. Yefeng Zheng Engineering & Technology Computer Science World Rank #24,697 Ming Xu. Researcher Publications. edu. Tencent’s Mr Tong was hired from Baidu Dong Yu View Recently neural network for singing voice synthesis [1,2,3,4] has attracted a lot of attention since deep learning has achieved great gain in text-to-speech (TTS) task [5,6,7] which Tencent AI Lab Seattle (tencent. Dong Yu, a distinguished scientist at Tencent who specializes in speech technologies and is an ACM/IEEE/ISCA Fellow. In this paper, we propose novel bilateral denoising diffusion models (BDDMs), which take signif- 8. com Abstract In this paper, we propose a Chinese multi-turn topic-driven conversation dataset, NaturalConv, which allows the partic-ipants to chat anything they want as long as any element Before joining StepFun, I worked as a research director at Tencent for four years and another five years at Megvii (Face++). Title. Yihao Liu. Before that, I received the Master's Degree from Tencent AI Lab, US. Interspeech, 2938-2942, 2018. He spent the past two decades at Microsoft Research and was part of a team that found a major breakthrough in Yu, who has been appointed as deputy head of Tencent's AI Lab division, will run the new lab as well as spearhead research in speech recognition and natural language understanding, the company said. com Dongyan Zhao Peking University Verified email at pku. May 2019 – August 2019 Bellevue, WA, USA. Advanced Data and Signal Processing Laboratory and School of Electronic and Computer Engineering, Peking University, Beijing, China, Dong Yu. arXiv. Industrial Invited Talk Talk 1 Title: Building Digital Human Presenter: Dong Yu, Tencent AI Lab Date: 13 October 2020, 16:30 - 17:00 (UTC) . We first discuss acoustic Experience: Littelfuse · Location: San Francisco Bay Area · 500+ connections on LinkedIn. Songxiang Liu, Shan Yang, Dan Su, Dong Yu Tencent AI Lab ABSTRACT Cross-speaker style transfer (CSST) in text-to-speech (TTS) synthesis aims at transferring a speaking style to the synthe-sised speech in a target speaker’s voice. NLP Research Intern Tencent AI Lab. 2022 Jiachen Lian, Chunlei Zhang, and Dong Yu. 66: 2018: Contribute to tencent-ailab/Lodoss development by creating an account on GitHub. Slides. Shi-Xiong (Austin) Zhang Sr. Direction: Image/Video Restoration, especially Super-Resolution Email: xintao. Affiliations: [Tencent AI Lab, Bellevue]. , 2022; Chen et al. Most previous CSST approaches rely on expensive high-quality data carrying de- Tencent Rhino-Bird Scholarship, Tencent Oct 2019 Hong Kong PhD Fellowship, HK Government 2018-Present Excellent Research Award, HKUST Sep 2018 Changshui Zhang, and Dong Yu MetaLogic: Logical Reasoning Explanations with Fine-Grained Structure. He has published more than 300 papers on topics including automatic speech Dong Yu (dongyu@ieee. , 2024). 160: 2024: CBLDNN-based speaker-independent speech separation via generative adversarial training. Linli Xu University of Science and Technology of China Verified email at ustc. AAAI 2021. Conference Papers: Xiyun Li, Yong Xu, Meng Yu, Shi-Xiong Zhang, Jiaming Xu, Bo Xu, Dong Yu. [13] proposed an approach to improve the MVDR beamformer for multichannel target speech separation by replacing the conventional covariance matrix estimation with a DNN Dong Yu Tencent AI Lab Bellevue WA, USA dyu@tencent. 2: 2022: The system can't perform the operation now. org) received his Ph. Most previous CSST methods in TTS require a reference utterance conveying the desired style to obtain Liqiang Zhang, Chengzhu Yu, Heng Lu, Chao Weng, Chunlei Zhang, Yusong Wu, Xiang Xie, Zijin Li, Dong Yu Tencent AI Lab Abstract. Tencent AI Lab, Bellevue, WA, USA, Yuexian Zou. ORCID provides an identifier for individuals to use with their name as they engage in research, scholarship, and innovation activities. PIVOINE: Instruction Tuning for Open-world Information Extraction. Researcher, Tencent AI Seattle Lab - Cited by 3,395 - LLM Agent - Natural Language Processing - Large Language Model Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. In this paper, we propose a multi-look neural network Yiwu Zhong 1, Liwei Wang 2, Jianshu Chen 2, Dong Yu 2, Yin Li 1. Publication Topics Bounding Box,Image Regions,Object Detection,Visual Grounding,Anchor Boxes,Candidate Regions,Contrastive Chengzhu Yu, Heng Lu, Na Hu, Meng Yu, Chao Weng, Kun Xu, Peng Liu, Deyi Tuo, Shiyin Kang, Guangzhi Lei, Dan Su, Dong Yu Tencent AI Lab ABSTRACT In this paper, we present a generic and robust multimodal synthesis system that produces highly natural speech and facial expression simultaneously. of Surrey, UK. Bosch Research, Palo Alto, CA. [pdf] [code] [bib] We present a generic and robust multimodal synthesis system that produces highly natural speech and facial expression simultaneously. Prior to joining Tencent America in 2017, he was a Dong Yu (dongYu@ieee. Read more. , author={Sangwoo Cho, Kaiqiang Song, Xiaoyang Wang, Fei Liu, and Dong Yu}, booktitle={Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing}, year Now a Principal Researcher in Tencent AI lab, Bellevue, USA. PIKA is a lightweight speech processing toolkit based on Pytorch and (Py)Kaldi. org. The Seattle AI Lab will be led by Dong Yu, who was recently poached from his role as a principal researcher at Microsoft, Tencent announced on Tuesday. qiao@siat. Chao Weng and Dr. The Chinese University of Hong Kong. Tencent AI Lab. I am a principal researcher at Tencent AI Platform Department (also known as the Game AI department of Tencent AI Lab). Yiwu Zhong, Liwei Wang, Jianshu Chen, Dong Yu, Yin Li. Considering existing challenges, the concept of self-correction and self-learning have been proposed as promising solutions (Madaan et al. She was also chosen by Southern Metropolis Daily as one of the "Four Dan actresses of the post-90s Generation", along with Zheng Shuang, Guan Xiaotong and Yang Zi. Sreyan Ghosh Ph. com} Abstract Large language models (LLMs) have been successfully adapted for interactive decision Renrui Zhang, Zhengkai Jiang, Ziyu Guo, Shilin Yan, Junting Pan, Hao Dong, Yu Qiao, Peng Gao, Hongsheng Li. ACL 2024-main 1, 1764–1775, 2024. Dong Yu (俞栋) Engineering & Technology Computer Science World Rank #11,226 Yefeng Zheng Tencent AI Lab. Supervisor: Songxiang Liu, Chao Weng, and Bo Wu Selected Publications (* denotes equal contributions. Dan Roth Professor of Zhou Dongyu (Chinese: 周冬雨; born 31 January 1992) is a Chinese actress, who gained recognition after appearing in Zhang Yimou's film Under the Hawthorn Tree. This work was partially developed when the first author was an intern at Tencent AI Lab, Bellevue, USA, and further completed at UW-Madison. & Tech. 35: 2021: Deformer: Towards displacement Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. His research works, which focus on statistical speech recognition and processing, have been recognized by the prestigious IEEE Signal Publications [35] Cognitive Kernel: An Open-source Agent System towards Generalist Autopilots Hongming Zhang, Xiaoman Pan, Hongwei Wang, Kaixin Ma, Wenhao Yu, Dong Yu. We are delighted to welcome Dr. Wenlin Yao Senior Scientist@Amazon | Ex Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. Try Kaiqiang Song (宋凯强) is a Senior Research Scientist of Natural Language Processing at Tencent AI Lab, Seattle. PRESS RELEASE PR Newswire . [pdf] [code] [bib] Xiaoyang Wang, Chen Li, Jianqiao Zhao, and Dong Yu. com Max W. arXiv preprint arXiv:2401. Tencent - Cited by 8,151 - Computer Vision - Deep Learning - Medical Imaging Analysis DONG WEI Tencent Verified email at tencent. 28: 2023: Learning Procedure-Aware Video Representation From Instructional Videos and Their Narrations. Xu Li1, Na Li 2, Chao Weng , Xunying Liu1, Dan Su2, Dong Yu2, Helen Meng1 1The Chinese University of Hong Kong 2Tencent AI Lab, Tencent fxuli, xyliu, hmmengg@se. com I am a PhD student at The Chinese University of Hongkong, majoring in Speech and Audio Processing, Supervised by Prof. We first discuss acoustic Manjie Xu, Chenxing Li, Yong Ren, Rilin Chen, Yu Gu, Wei Liang, Dong Yu Tencent AI Lab. Verified email at cse. Signal Processing Digital Library* Dr. Interspeech, 2262-2266, 2018. Try again later. xu. D Su, L He, L Xu, D Yu. Sc. Dong Yu is heading up the new lab for Tencent. Proceedings of the 22nd ACM SIGKDD international conference on knowledge Technical Report Fortunately, personas are very easy to scale up. MDT-A2G: Exploring Masked Diffusion Transformers for Co-Speech Gesture Generation. He is currently taking the role of a distinguished scientist and vice general manager at Tencent AI Lab. Zebra: Extending Context Window with Layerwise Grouped Local-Global Attention Kaiqiang Song∗, Xiaoyang Wang∗, Sangwoo Cho Hao Zhang, Meng Yu, Dong Yu Tencent AI Lab, Bellevue, WA, USA ABSTRACT In this paper, we formulate acoustic howling suppression (AHS) as a supervised learning problem and propose a deep learning ap-proach, called Deep AHS, to address it. Interspeech 2022 . Prior to joining Tencent in 2017, he worked as a principal Dong Yu is a Distinguished Scientist and Vice General Manager at Tencent AI Lab, which has teams in Shenzhen, Beijing, and Seattle. In this paper, we propose a Unified Speech Enhancement and Editing (uSee) model with conditional diffusion models to handle various tasks at the same time in a generative manner. The series broke 190 million views on the first day of release and 1 billion views in just a week. Chinese stars Xu Kai and Zhou Dong-yu charmed the hearts of many in the 2021 drama 'Ancient Love Poetry. Cross-speaker style transfer (CSST) in neural text-to-speech (TTS) synthesis aims at transferring a speaking style to the synthesised speech in a target speaker's voice. SHENZHEN, China, May 2, 2017 /PRNewswire/ -- Tencent * Dec 14, 2024 According to Total H Index. com. Meng YU Tencent AI Lab Verified email at tencent. Executing your Commands via Motion Diffusion in Latent Space Meng Yu(Tencent AI Lab), Xuan Ji(Tencent AI Lab), Bo Wu(Tencent AI Lab), Dan Su(Tencent AI Lab) and Dong Yu(Tencent AI Lab) Abstract: The performance of keyword spotting (KWS), measured in false alarms and false rejects, degrades significantly under the far field and noisy conditions. InterSpeech 2021, Brno, Czechia. Yuan Yu Xuan as Cheng Yu; An immortal who is brought from the mortal realm to the Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. Yu Dong as AI Lab Deputy Director and Opens New AI Lab in Seattle. 2, 2017, 05:05 AM. Articles 1 Dong Yu and Jinyu Li Tencent AI Lab, USA. Wenhao Yu. com Abstract Denoising diffusion probabilistic models (DDPMs) have emerged as competitive generative models yet brought challenges to efficient sampling. Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow - Cited by 67,496 - automatic speech recognition - speech processing - natural language processing Loading Dr. Investigating Prior Knowledge for Challenging Chinese Machine Reading Comprehension. Y Ji, D Yu, MR Lyu, L Wang. arXiv preprint arXiv:2406. Zhengyou Zhang Engineering & Technology Computer Science World Rank #6,509 Dong Yu (俞栋) Dong Yu (俞栋) Engineering & Technology Computer Science World Rank #11,226 Yefeng Zheng. Xiaoyang Wang*, Chen Li*, Jianqiao Zhao, Dong Yu Tencent AI Lab, Bellevue, WA {shawnxywang,ailabchenli, markjzhao,dyu}@tencent. Tencent AI Lab Bothell, WA, USA. Could Small Language Models Serve as Recommenders? Toward Data-centric Cold-start Recommendation Xuansheng Wu, Huachi Zhou, Wenlin Yao, Xiao Huang Songxiang Liu, Dan Su, Dong Yu, DiffGAN-TTS: High-Fidelity and Efficient Text-to-Speech with Denoising Diffusion GANs, 2022 International Conference on Machine Learning Tencent AI Lab (09/2019-12/2022): Research Intern, conducting research Semantic Scholar profile for Dian Yu, with 209 highly influential citations and 39 scientific research papers. md at main · tencent-ailab/3m-asr Meng Yu, Yong Xu, Chunlei Zhang, Shi-Xiong Zhang, Dong Yu Tencent AI Lab, Bellevue, WA, USA fraymondmyu, lucayongxu, cleizhang, auszhang, dyug@tencent. Members: Free IEEE Members: $11. Outstanding Staff Award at Tencent, 2021, 2022, 2023; Tencent Technology Breakthrough Award, 2020, 2022; Neural Spatio-Temporal Beamformer for Target Speech Separation. The first release focuses on end-to-end speech recognition. We sat down with Yu to learn more about Tencent and its AI strategy. Conversational Semantic Role Labeling. "MIMO Self-attentive RNN Beamformer for Multi-speaker Speech Separation". Yu Dong is leading Tencent’s new A. ECCV 2020 Paper I am currently a Principal Researcher (supervised by Dr. J Yu, SX Zili Huang 1†, Yiwen Shao , Shi-Xiong Zhang 2, Dong Yu 1Center for Language and Speech Processing, Johns Hopkins University, Baltimore, MD, USA 2Tencent AI Lab, Bellevue, WA, USA ABSTRACT The speech field is evolving to solve more challenging scenar-ios, such as multi-channel recordings with multiple simultaneous talkers. Articles Cited by Public access Co-authors. Sentences produced by abstractive summarization systems can be ungrammatical and fail to preserve the original meanings, despite being locally fluent. Yifan Gong Principal Science Manager, Microsoft Corp. dongyu@ieee. However, singing data for target speaker is much more difficult to collect Dong Yu Tencent AI Lab Bellevue WA, USA dyu@tencent. 9K conversations from six domains, and 400K utterances with an average turn number of 20. com ABSTRACT Existing approaches for replay and synthetic speech detection still lack generalizability to unseen spoofing Dong Yu Tencent AI Lab Bellevue WA, USA dyu@tencent. 3M: Multi-loss, Multi-path and Multi-level Neural Networks for speech recognition - 3m-asr/README. Yu, and are confident that his addition to the team will serve to further strengthen Tencent's AI Kai Sun, Dian Yu, Dong Yu, and Claire Cardie. 2022. Verified email at tencent. Singing voice conversion is converting the timbre in the source singing to the target speaker's voice while keeping singing content the same. Although text-guided diffusion models have made significant advancements in text ORCID record for Zhengyu Dong. com), Zhuohuang Zhang, Meng Yu, Shi-Xiong Zhang, Dong Yu, Tencent AI lab, Bellevue, WA, USA Some tricks to successfully train a robust RNN beamformer: Meng Yu's 81 research works with 1,828 citations and 11,960 reads, including: Zoneformer: On-device Neural Beamformer For In-car Multi-zone Speech Separation, Enhancement and Echo Cancellation Dong Yu Tencent AI Lab Bellevue WA, USA dyu@tencent. Tencent. degree in computer science from the University of Idaho, an M. C Yuan, K Ma, S Yu, D Wei, Y Zheng. pdf •25 years global enterprise experience in AI, IT, cloud, SaaS, enterprise software · Experience: Tencent · Education: University of Wisconsin-Milwaukee · Location: San Jose · 500 . BETA VERSION. 06633, 2022. Subject field ranking in world, region, country and university is beta version since subject field 'others' (873,898) scientists profile whose branch cannot be determined, not yet edited or determined) is excluded, so the ranking will change as the others fields are edited. Daniel S. Rongzhi Gu Tencent AI Lab Verified email at pku. Abstract: Digital human finds its applications in areas such as virtual companion, virtual reporter, and virtual narrator. && Ph. 00. Publication Topics. IEEE Transactions on Medical Imaging 41 (5), 1043-1056, 2021. Professor Shanghai AI Lab, SIAT. com Abstract In this paper, we propose a Chinese multi-turn topic-driven conversation dataset, NaturalConv, which allows the partic-ipants to chat anything they want as long as any element Abstract. Dong Yu is an ACM/IEEE/ISCA Fellow. Y Zhong, L Yu, Y Semantic Scholar profile for Dian Yu, with 209 highly influential citations and 39 scientific research papers. Dong Yu on multi-source localization and its importance for multi-talker speech recognition. [3] In 2016, she won the Golden Horse Liqiang He 1, Shulin Feng , Dan Su , Dong Yu2 1Tencent AI Lab, Shenzhen, China 2Tencent AI Lab, Bellevue WA, USA ABSTRACT Neural architecture search (NAS) has attracted much attention and has been explored for automatic speech recognition (ASR). org) received a Ph. He works at Tencent AI Lab as a distinguished scientist and vice general manager. TASLP. , PMP’s profile on LinkedIn, a professional community of 1 billion members. Smart assistant: The HunyuanAide project team will work together with other internal teams at Tencent to develop a large language model using reinforcement learning algorithms for training. Interspeech2024, 2024. In this work, we focus on streaming ASR scenarios and propose the latency-controlled NAS for Chengqi Deng 1, Chengzhu Yu 2, Heng Lu 2, Chao Weng 2, Dong Yu 2. Bo Wu Senior Researcher, Tencent AI Lab, Shenzhen, China Verified email at tencent. Received: September 01 2019. Dong Yu The task of few-shot style transfer for voice cloning in text-to-speech (TTS) synthesis aims at transferring speaking styles of an arbitrary source speaker to a target speaker's voice Considering existing challenges, the concept of self-correction and self-learning have been proposed as promising solutions (Madaan et al. News. Jianwei Yu, Yan Wang*, Nuo Chen, Yanyao Bian, Guangzhi Li, Deng Cai, Dong Yu. View Dong Yu, Ph. On ArXiv. To submit a bug report or feature request, you can use the official OpenReview GitHub repository: Report an issue Dong Yu is an IEEE Fellow, ISCA Fellow, and an ACM distinguished scientist. degree from the University of Science and Technology of China (USTC) in 2015 and studied Meng YU Tencent AI Lab Verified email at tencent. We introduce sub-sentence encoder, a contrastively-learned contextual embedding model for fine-grained semantic representation of text. In this paper we propose to remedy this problem by jointly generating a sentence and its syntactic dependency parse while performing abstraction. cn Xintao Wang. This is a very chal- Chao Dong. [ paper] Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. [ paper] [34] Abstraction-of-Thought Makes Language Models Better Reasoners. 2020. hk, lina011779@126. Deep AHS is trained in a teacher forcing way which converts the recurrent howling suppres- 俞栋, Distinguished Scientist, Tencent AI Lab, Dr. Specifically, by providing multiple types of conditions including self-supervised learning embeddings and proper text prompts to the score-based diffusion model, we can enable Tencent AI Labs - Cited by 176 - Microphone Array Processing - Machine Learning - Speech Enhancement - Distant Speech Recognition Meng YU Tencent AI Lab Verified email at tencent. Dong Yu has concentrated his research on speech recognition and processing, as well as natural language processing in recent years. Within these framework, LLMs typically operate by employing two main strategies: 1) they continuously refine their responses based on the feedback of their past responses, and 2) they Manjie Xu, Chenxing Li, Yong Ren, Rilin Chen, Yu Gu, Wei Liang, Dong Yu Tencent AI Lab. Prior to joining Tencent in 2017, he was a principal researcher at Microsoft speech and dialog research group. Peter Clark Allen Institute for Artificial Intelligence (AI2) Verified email at allenai. 2005 Harvard University Physics Postdoc 2005-2008 I had a wonderful time in the industry as a research intern: Meta AI in 2022, Microsoft Research in 2021, Tencent AI Lab-Seattle in 2019, and Tencent AI Lab-Shenzhen in 2018. From massive web data, we automatically construct Persona Hub — a persona collection containing 1 billion diverse personas (∼13% of the world’s total MM-LLMs: Recent Advances in MultiModal Large Language Models Duzhen Zhang1*‡, Yahan Yu3*, Jiahua Dong4†, Chenxing Li1, Dan Su1, Chenhui Chu3† and Dong Yu2 1Tencent AI Lab, China 2Tencent AI Lab, USA 3Kyoto University, Japan 4Mohamed bin Zayed University of Artificial Intelligence, United Arab Emirates {duzhen. This is a very chal- Tencent Holdings Ltd. Researcher, Tencent AI Seattle Lab. S. Wenlin Yao Senior Scientist@Amazon | Ex Citation @article{hifi3dface2021tencentailab, title={High-Fidelity 3D Digital Human Head Creation from RGB-D Selfies}, author={Bao, Linchao and Lin, Xiangkai and Chen, Yajing and Zhang, Haoxian and Wang, Sheng and Zhe, Xuefei and Kang, Di and Huang, Haozhi and Jiang, Xinwei and Wang, Jue and Yu, Dong and Zhang, Zhengyou}, journal={ACM Transactions on Songxiang Liu, Shan Yang, Dan Su, Dong Yu 2 Tencent AI Lab Introduction. 2024: The system can't perform the operation now. Yu, and are confident that his addition to the team will serve to further strengthen Tencent's AI 俞栋，语音识别与深度学习领域的专家。现任腾讯AI Lab（人工智能实验室）副主任。 Top 10 Scientists in Tencent. [paper] [code] Tencent AI Lab, Bellevue, WA. Tencent AI Lab; CASIA - Cited by 299 - Far-field Speech Recognition - Speech Enhancement - Speech Separation D Zhang, Y Yu, J Dong, C Li, D Su, C Chu, D Yu. Properties of CTC 1/21/18 Dong Yu : State-of-the-art of End-to-end Speech Recognition Systems 8 CTC (Spiky): output blanks until confident to output the associated label • Simple: direct audio sequence to label sequence transformation, flexible modeling unit • Fast: decoding is fast due to spikes (confident output) or less output units • Random timing: spikes can happen Tencent AI lab - Cited by 1,885 - ASR J Yu, SX Zhang, J Wu, S Ghorbani, B Wu, S Kang, S Liu, X Liu, H Meng, ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Principal Researcher, Tencent America, Bellevue, USA - Cited by 6,712 - Neural Beamforming - ASR - Multi-modal LLM - Speech Enhancement - Sound Event Detection Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. D degree in computer science from the University of Idaho. 82: 2018: Enhancing end-to-end multi-channel speech separation via spatial feature learning. Verified email at microsoft. Chunlei Zhang Neural Spatio-Temporal Beamformer for Target Speech Separation. State-of-the-art methods for STS task use language models Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. "Dr. Once a Research Fellow at Uni. I. G Dong, K Lu, C Li, T Xia, B Yu, C Zhou, J Zhou. In contrast to the standard practice with sentence embeddings, where the meaning of an entire sequence of text is encoded into a fixed-length vector, the sub-sentence encoder learns to produce distinct contextual embeddings Tencent AI Lab, US Publication Topics Absence Of Context,Analysis System,Classification System,Commonsense Knowledge,Comprehensive Benchmark,Description Of The Relationship,Free Base,Graph Structure,Head And Tail,Linguistic Knowledge,Missing Link,Promising Future Directions,Properties Of Translation,Textual Descriptions,Training Dong Yu The task of few-shot style transfer for voice cloning in text-to-speech (TTS) synthesis aims at transferring speaking styles of an arbitrary source speaker to a target speaker's voice Tencent AI lab - Cited by 1,885 - ASR J Yu, SX Zhang, J Wu, S Ghorbani, B Wu, S Kang, S Liu, X Liu, H Meng, ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Tencent AI Lab Bothell, WA, USA. Wenwu Wang) for two years. Manjie Xu, Chenxing Li*, Duzhen zhang, Dan Su, Wei Liang*, Dong Yu* *corresponding author. In EMNLP 2023 (Findings). Zhaowei Wang ORCID record for Zhengyu Dong. The series airs on Tencent Video starting January 22, 2020. Dan Roth Professor of Zhang will be assisted by Dr. 2 Tencent AI Lab, Bellevue, USA. degree in physics from Nanjing University, Nanjing, China, in 2017 an We present a fully automatic system that can produce high-fidelity, photo-realistic three-dimensional (3D) digital human heads with a consumer RGB-D selfie camera. 00 Non-members: $15. S. Publication Topics Bounding Box,Image Regions,Object Detection,Visual Grounding,Anchor Boxes,Candidate Regions,Contrastive Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. Yu Dong is an expert on speech recognition and deep learning. com Abstract—In this paper, we summarize recent progresses made in deep learning based acoustic models and the motivation and insights behind the surveyed techniques. Dan Su Tencent AI Lab Verified email at tencent. Songxiang Liu, Dan Su, Dong Yu Tencent AI Lab ABSTRACT The task of few-shot style transfer for voice cloning in text-to-speech (TTS) synthesis aims at transferring speaking styles of an arbitrary source speaker to a target speaker’s voice using very limited amount of neutral data. Dong Yu has been appointed as deputy director of Tencent Holding’s new AI Lab in Seattle. com), Meng Yu, Shixiong Zhang, Lianwu Chen, Chao Weng, Jianming Liu, Dong Yu, Tencent AI lab, Bellevue, WA, USA Purely NN based speech separation and enhancement methods, although can achieve good objective scores, inevitably Part of this work was conducted when K. Recent work shows that unsupervised singing voice conversion can be achieved with an autoencoder-based approach My name is WANG Yan (王琰). Prior to joining Tencent in 2017, he worked as a principal researcher at Microsoft Research (Redmond), where he had been since 1998. tencent. His research works, which Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. 2015. Generalized Spatio-Temporal RNN Beamformer for Target Speech Separation. Refer to google scholar find all Publications) Dongchao Yang, Jianwei Yu, Helin Wang, Wen Wang, Chao Weng, Yuexian Zou, Dong Yu Diffsound: Discrete Diffusion Model for Text-to-sound generation Yu, who has been appointed as deputy head of Tencent's AI Lab division, will run the new lab as well as spearhead research in speech recognition and natural language understanding, the company said. Properties of CTC 1/21/18 Dong Yu : State-of-the-art of End-to-end Speech Recognition Systems 8 CTC (Spiky): output blanks until confident to output the associated label • Simple: direct audio sequence to label sequence transformation, flexible modeling unit • Fast: decoding is fast due to spikes (confident output) or less output units • Random timing: spikes can happen Na Li's 16 research works with 240 citations and 1,201 reads, including: The DKU-Tencent System for the VoxCeleb Speaker Recognition Challenge 2022 I am currently a senior research scientist at Tencent AI Lab, Seattle. Dong Yu) at Tencent America · Experience: Tencent Americas · Education: Georgia Institute of Technology · Location: Bellevue · 500 Principal Researcher, Tencent America, Bellevue, USA - Cited by 6,726 - Neural Beamforming - ASR - Multi-modal LLM - Speech Enhancement - Sound Event Detection Yiwen Shao, Shi-Xiong Zhang, Yong Xu, Dong Yu, Daniel Povey, Sanjeev Khudanpur. NaturalConv: A Chinese Dialogue Dataset Towards Multi-Turn Topic-Driven Conversation. Dong Yu; This paper proposes a new paradigm for handling far-field multi-speaker data Zhou Dongyu (Chinese: 周冬雨; born 31 January 1992) is a Chinese actress, who gained recognition after appearing in Zhang Yimou's film Under the Hawthorn Tree. Before joining Tencent America in 2017, he Enter your feedback below and we'll get back to you as soon as possible. May. Medical Image Analysis. Dong Yu) at Tencent America (AI Lab), Bellevue, WA, USA. Hongwei Wang and Dong Yu Tencent AI Lab, Seattle, WA {hongweiw, dyu}@global. We aim to offer insights into the video-to Dong Yu View Recently neural network for singing voice synthesis [1,2,3,4] has attracted a lot of attention since deep learning has achieved great gain in text-to-speech (TTS) task [5,6,7] which Industrial Invited Talk Talk 1 Title: Building Digital Human Presenter: Dong Yu, Tencent AI Lab Date: 13 October 2020, 16:30 - 17:00 (UTC) . Recent work proposed advanced prompting techniques and the necessity of fine-tuning with high-quality data to augment LLMs' reasoning abilities. com Abstract Acoustic echo cancellation (AEC) plays an important role in the full-duplex speech communication as well as the front-end speech enhancement for recognition in the conditions when the Internship | Tencent AI Lab (Seattle) Advisor: Dian Yu and Dong Yu Feb 2024 - Apr 2024, Bellevue, Washington, United States. hk. D Yu, A Tencent. Kun Li Hao Zhang, Meng Yu, Dong Yu Tencent AI Lab, Bellevue, WA, USA ABSTRACT In this paper, we formulate acoustic howling suppression (AHS) as a supervised learning problem and propose a deep learning ap-proach, called Deep AHS, to address it. Dong Yu is an IEEE Fellow, an ISCA Fellow, and an ACM distinguished scientist. Kaixin Ma Hongming Zhang Hongwei Wang Xiaoman Pan Wenhao Yu Dong Yu Tencent AI Lab, Bellevue, WA {kaixinma,hongmingzhang,hongweiw,xiaomanpan,wenhaowyu,dyu@global. IJCAI, 2022(oral). See all Zhengyou Zhang. ustc@gmail. ' Adapted from the novel, “Ancient God”, by Xing Ling, “Ancient Love Poetry” is a 2021 historical fantasy romance drama directed by Yin Tao and Li Cai. com Xiaoyang Wang*, Chen Li*, Jianqiao Zhao, Dong Yu Tencent AI Lab, Bellevue, WA {shawnxywang,ailabchenli, markjzhao,dyu}@tencent. 2017 - present. Quoc V. Cognitive Kernel: An Open-source Agent System towards Generalist Autopilots. 2023 – 2024 Coworker. 2000 University of Chicago Physics Ph. Dian Yu Tencent AI Lab Verified email at Dong Yu's 356 research works with 8,208 citations and 2,844 reads, including: Discover, Explain, Improve: An Automatic Slice Detection Benchmark for Natural Language Processing Lifeng Jin, Baolin Peng, Linfeng Song, Haitao Mi, Ye Tian and Dong Yu Tencent AI Lab, Bellevue, WA lifengjin@global. We propose a new bilateral denoising diffusion model (BDDM) that parameterizes both the forward and reverse processes with a schedule network 俞栋，语音识别与深度学习领域的专家。现任腾讯AI Lab（人工智能实验室）副主任。 "Dr. Contribute to tencent-ailab/Lodoss development by creating an account on GitHub. com ABSTRACT Inthiswork,threelattice-free(LF)discriminativetraining criteria for purely sequence-trained neural network acoustic models are compared on LVCSR tasks, namely maximum mutual information (MMI), boosted maximum mutual infor-mation(bMMI)andstate-levelminimumBayesrisk Senior Researcher, Tencent AI Lab, Shenzhen, China - Cited by 495 - speech enhancement - acoustic modeling - microphone array signal processing DM Zhenyu Tang, Lianwu Chen, Bo Wu, Dong Yu. Within these framework, LLMs typically operate by employing two main strategies: 1) they continuously refine their responses based on the feedback of their past responses, and 2) they Senior Scientist@Amazon | Ex-Tencent AI Lab - Cited by 1,052 - Natural Language Processing - Computational Linguistic - Machine Learning Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. Coworker. 110: Affiliations: [Tencent AI Lab, Bellevue]. Revision Received: December 01 2019 Kai Sun, Dian Yu, Dong Yu, Claire Cardie; Investigating Prior Knowledge for Challenging Chinese Machine Reading Comprehension. [3] In 2016, she won the Golden Horse Speech and NLP Group, Tencent AI Lab(American) , WA, USA Research Intern • Dec 2021 to Feb. Beijing Institute of Technology, Tencent AI Lab Beijing, Tencent AI Lab Seattle. Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. com Abstract In this paper, we propose a Chinese multi-turn topic-driven conversation dataset, NaturalConv, which allows the partic-ipants to chat anything they want as long as any element Songxiang Liu, Dan Su, Dong Yu Tencent AI Lab ABSTRACT The task of few-shot style transfer for voice cloning in text-to-speech (TTS) synthesis aims at transferring speaking styles of an arbitrary source speaker to a target speaker’s voice using very limited amount of neutral data. Lifa Sun Co-founder, DailyLive Limited Verified email at dailylive. Accepted to Interspeech2021, *Yong XU (yong. com Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. of China (USTC) Materials Science B. , Microsoft AI and Research, USA. Y. Le Research Scientist, Google Verified email at stanford. I will serve as an Chi Zhang, Wei Yin, Gang Yu, Zhibin Wang, Tao Chen, Bin Fu, Tianyi Zhou, Chunhua Shen ICCV, 2023. Worked with Dr. This is a very chal- Dong Yu View Zhang et al. D Yu, JC Mao. Outstanding Staff Award at Tencent, 2021, 2022, 2023; Tencent Technology Breakthrough Award, 2020, 2022; Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. Tencent - Cited by 1,152 - Medical Image Analysis DONG WEI. He has two monographs and over 300 papers to his Top 10 Scientists in Tencent. Cheng Bian Bytedance Verified email at bytedance. Dong Yu and Jinyu Li Tencent AI Lab, USA. has hired Yu Dong (pictured), a speech recognition expert and former principal scientist at Microsoft Research, as its vice general AI Dong Yu Tencent AI Lab Bellevue WA, USA dyu@tencent. 13542, 2024. Try Despite the impressive capabilities of Large Language Models (LLMs) on various tasks, they still struggle with scenarios that involves complex reasoning and planning. NAACL (oral), 2024. arXiv preprint arXiv:2202. com Abstract Retrieval-augmented language model (RALM) represents a significant advancement in miti-gating factual hallucination by leveraging ex-ternal knowledge sources. org, jinyli@microsoft. in CS at University of Maryland, College Park Verified email at umd. Senior Researcher, Tencent AI Lab Seattle. Shinji Watanabe Carnegie Mellon University Verified email at cmu. Code. degree in physics from Nanjing University, Nanjing, China, in 2017 an Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. com, fcweng, dansu, dyug@tencent. com) 2018 – 2021 Suggest Position; Advisors, Relations & Conflicts. Generating semantically and temporally aligned audio content in accordance with video input has become a focal point for researchers, particularly following the remarkable breakthrough in text-to-video generation. The system only needs the user to take a short selfie RGB-D video while rotating his/her Songxiang Liu, Dan Su, Dong Yu Tencent AI Lab ABSTRACT The task of few-shot style transfer for voice cloning in text-to-speech (TTS) synthesis aims at transferring speaking styles of an arbitrary source speaker to a target speaker’s voice using very limited amount of neutral data. Plumbley and Prof. In this work, we focus on streaming ASR scenarios and propose the latency-controlled NAS for SEED, Bytedance; Ex-Tencent AI Lab - Cited by 1,600 - speech processing - ASR - TTS - speaker recognition - machine learning Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. com Abstract In this paper, we propose a Chinese multi-turn topic-driven conversation dataset, NaturalConv, which allows the partic-ipants to chat anything they want as long as any element Dong Yu — disambiguation page; Dong Yu 0002 — Xi'an Jiaotong University, Institution of Advanced Manufacturing and Technology, China; Dong Yu 0003 — Beijing Language and Culture University, Beijing, China; Dong Yu 0004 — University of Chinese Academy of Sciences, Beijing, China (and 2 more) Vinay Kothapally (Tencent AI); Yong Xu (Tencent); Meng Yu (Tencent); Shixiong Zhang (Tencent); Dong Yu (tencent) SPS. com} Abstract Large language models (LLMs) have been successfully adapted for interactive decision Xiaoyang Wang*, Chen Li*, Jianqiao Zhao, Dong Yu Tencent AI Lab, Bellevue, WA fshawnxywang, ailabchenli, markjzhao, dyug@tencent. was an intern at the Tencent AI Lab, Bellevue, WA. [CODE & DATA] Dong Yu In this paper, we propose a novel unsupervised text-to-speech (UTTS) framework which does not require text-audio pairs for the TTS acoustic modeling (AM). This is a very chal- Short Bio: Dr. 13: 2024: Multi-hop Evidence Retrieval for I had a wonderful time in the industry as a research intern: Meta AI in 2022, Microsoft Research in 2021, Tencent AI Lab-Seattle in 2019, and Tencent AI Lab-Shenzhen in 2018. Tencent Appoints Dr. Before that, Wenhao Yu, and Dong Yu. We use Pytorch as deep learning engine, Kaldi for data formatting and feature extraction. Conference on Empirical Methods in Natural Language Processing (EMNLP). Lam, Jun Wang, Dan Su, Dong Yu. Researcher Tencent. 1 Zhejiang University, 2 Tencent AI Lab. ECCV 2020, Glasgow, UK. Park Google Brain Verified email at google. He is a distinguished scientist and vice general manager at Tencent AI Lab, Bothell, Short Bio: Dr. Deep AHS is trained in a teacher forcing way which converts the recurrent howling suppres- Tencent AI Lab, Speech Group, Intern. Dong Yu. I earned my Ph. In this talk, I will introduce our recent work in building digital human. com Abstract Semantic Textual Similarity (STS) measures the degree to which the underlying semantics of paired sentences are equivalent. Direction: Image/Video Restoration, Super-Resolution Email: yu. Chunlei Zhang Dr. Rongjie Huang, Zhou Zhao, Huadai Liu, Jinglin Liu, and Principal Researcher, Tencent America, Bellevue, USA - Cited by 6,712 - Neural Beamforming - ASR - Multi-modal LLM - Speech Enhancement - Sound Event Detection Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. com XPixel at Shanghai AI Lab. Top Reasons to Join SPS Today! 1. In this paper, we propose novel bilateral denoising diffusion models (BDDMs), which take signif- Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. R Gu, SX Zhang, L Chen, Y Xu, M Yu, D Su, Y Zou, D Yu. Follow. Our corpus contains 19. 03182, 2023. Mark D. Affiliations: [Tencent AI Lab, Shenzhen, China]. , HKUST - Cited by 3,980 - Natural Language Processing - World Knowledge Representation - Reasoning with LLM Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Fellow Verified email at global. cn Zhiguo Wang Principal Scientist at AWS AI Labs Verified email at amazon. 2022 ICASSP: Masked Proxy Loss For Text-Independent Speaker Verification Jiachen Lian, Aiswarya Vinod Kumar , Hira Dhamyal, Bhiksha Raj, and Chengzhu Yu,Heng Lu,Na Hu,Meng Yu,Chao Weng,Kun Xu,Peng Liu,Deyi Tuo,Shiyin Kang,Guangzhi Lei,Dan Su,Dong Yu +11 more Tencent,The Chinese University of Hong Kong,Beijing University of Posts and Telecommunications - Semantic Scholar profile for Dong Yu, with 2426 highly influential citations and 573 scientific research papers. Helen Meng. cn. Lianwu CHEN Kuaishou Technology Verified email at kuaishou. Technical Report. Jianqiao Zhao. Tencent AI Lab, Bellevue, WA, USA Dong Yu View A speaker counting technique has been proposed using a scheme that alternates between speech enhancement Page 2 of 21 Hsu and Bai EURASIP Journal on Audio, Speech, and Music Liqiang He 1, Shulin Feng , Dan Su , Dong Yu2 1Tencent AI Lab, Shenzhen, China 2Tencent AI Lab, Bellevue WA, USA ABSTRACT Neural architecture search (NAS) has attracted much attention and has been explored for automatic speech recognition (ASR). com Dr. I once worked at the University of Surrey, UK as a Research Fellow (supervised by Prof. The key component of this system is the Duration I am a senior research scientist at Tencent AI Lab. His research interests center on advancing artificial intelligence (AI) Kaiqiang Song, Hongming Zhang, Dong Yu, Jianshu Chen (2023). Google Scholar / CV / Zhihu. Singing voice conversion is to convert a singer's voice to another one's voice without changing singing content. Yao-Jen (Kevin) S Yu, K Ma, Q Bi, C Bian, M Ning, N He, Y Li, H Liu, Y Zheng. Most previous CSST approaches rely on expensive high-quality data carrying de- * Dec 14, 2024 According to Total H Index. submitted to Interspeech2020, Yong XU (yong. degree in computer science from Indiana University at Bloomington, an M. (TACL). of Sci. ICASSP 2020, 2020. He is close with Dong Hua and Lian Song, as well as Feng Jiu whom he helped get close to Dong Hua. Introduction. Vienna, Austria. Audio editing involves the arbitrary manipulation of audio content through precise control. Research intern May 2015 - Aug. cuhk. Abstract: Diffusion probabilistic models (DPMs) and their extensions have emerged as competitive generative models yet confront challenges of efficient sampling. DONG YU Professional Preparation: Univ. Internship | Tencent AI Lab (Seattle) Advisor: Dian Yu and Dong Yu May 2023 - Aug 2023, Bellevue, Washington, United States. zhang972,dongjiahua1995}@gmail. The aim SPS member Dr. 42 * 2020: Audio-Visual Multi-Channel Integration and Recognition of Overlapped Speech. 13601, 2024. , 2024; Saunders et al. Dong Yu, IEEE Fellow, is a distinguished scientist and vice general manager at Tencent AI Lab. Automatic Prosody Annotation with Pre-Trained Text-Speech Model. My research focuses on the areas of summarization, long document understanding, and understanding of LLMs. Ruixin Hong, Hongming Zhang, Xiaoman Pan, Dong Yu, Changshui Zhang. com), Meng Yu, Shixiong Zhang, Lianwu Chen, Chao Weng, Jianming Liu, Dong Yu, Tencent AI lab, Bellevue, WA, USA Purely NN based speech separation and enhancement methods, although can achieve good objective scores, inevitably 3M: Multi-loss, Multi-path and Multi-level Neural Networks for speech recognition - tencent-ailab/3m-asr 8. 1. Senior researcher Nov. Internship | Allen Institute (AI2) Advisor: Ashwin Kalyan and Peter Clark Aristo Research Team Kaixin Ma Hongming Zhang Hongwei Wang Xiaoman Pan Wenhao Yu Dong Yu Tencent AI Lab, Bellevue, WA {kaixinma,hongmingzhang,hongweiw,xiaomanpan,wenhaowyu,dyu@global. Author Bio: Jianwei Yu received the B. He currently works at Tencent AI Lab as a distinguished scientist and vice general manager. We aim to offer insights into the video-to Renrui Zhang, Zhengkai Jiang, Ziyu Guo, Shilin Yan, Junting Pan, Hao Dong, Yu Qiao, Peng Gao, Hongsheng Li. ac. ProDiff: Progressive Fast Diffusion Model for High-Quality Text-to-Speech. ECCV 2020 Paper I was grateful to intern or collaborate at TikTok, Shanghai AI Lab, Tencent Seattle Lab, Alibaba Qwen, with Yi Ren, Jinglin Liu, Chunlei Zhang Dong Yu, Yi Ren, and Zhou Zhao. In this paper, we propose novel bilateral denoising diffusion models (BDDMs), which take signif- May, 2023: I joined the Tencent AI Lab (@Seattle) as a research intern! Apr, 2023: One paper is accepted by AIED 2023! Xiaoman Pan, Xiaoyang Wang, Ninghao Liu, and Dong Yu. Signal Processing Digital Library* Dong Yu's 356 research works with 8,208 citations and 2,844 reads, including: Discover, Explain, Improve: An Automatic Slice Detection Benchmark for Natural Language Processing Now I am a Principal Research Scientist (supervised by Dr. Deng Cai, Yan Wang*, Huayang Semantic Scholar profile for Dong Yu, with 2426 highly influential citations and 573 scientific research papers. C Yu, H Lu, N Hu, M Yu, C Weng, K Xu, P Liu, D Tuo, S Kang, G Lei, D Su, Interspeech, 2027-2031, 2020. Immortal Shang Gu (Zhou Dong-yu) has spent many lifetimes as the leader of the four true immortals of Xiaoyang Wang*, Chen Li*, Jianqiao Zhao, Dong Yu Tencent AI Lab, Bellevue, WA fshawnxywang, ailabchenli, markjzhao, dyug@tencent. In this paper, we propose novel bilateral denoising diffusion models (BDDMs), which take signif- Distinguished Scientist at Tencent, Director of ARC Lab & AI Lab CVC - Cited by 16,427 - Deep learning - computer vision - machine learning - paid search - display ads X Wang, L Xie, C Dong, Y Shan. 07 Jun 2023 While current deep learning (DL)-based beamforming techniques have been proved effective in speech separation, they are often designed to process narrow-band Affiliations: [Tencent AI Lab, Shenzhen, China]. ICLR 2024 / Paper / Code. Most previous CSST approaches rely on expensive high-quality data carrying de- Meng YU Tencent AI Lab Verified email at tencent. com Abstract The most common training pipeline for large language models includes pretraining, finetun-ing and aligning phases, with their respective resulting models, such as the pretrained model and the finetuned model. Kun Xu, Han Wu, Linfeng Song, Haisong Zhang, Linqi Song and Dong Yu. However, the reli- Dr. 1 University of Wisconsin-Madison, USA. D. Director Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, ACM/IEEE/ISCA Dong Yu (俞栋) Distinguished Scientist @ Tencent AI Lab, P Karypis, Z Wang, C Dong, A Gentili, arXiv preprint arXiv:2310. If generating a word can introduce an erroneous relation to the Songxiang Liu, Shan Yang, Dan Su, Dong Yu Tencent AI Lab ABSTRACT Cross-speaker style transfer (CSST) in text-to-speech (TTS) synthesis aims at transferring a speaking style to the synthe-sised speech in a target speaker’s voice. Kaiqiang Song, Chao Zhao, Xiaoyang Wang, Dong Yu arXiv. However, these approaches are inherently SPATIALCODEC: NEURAL SPATIAL SPEECH CODING Zhongweiyang Xu⋆, Yong Xu †, Vinay Kothapally , Heming Wang ♯, Muqiao Yang ‡, Dong Yu † †Tencent AI Lab, ⋆ University of Illinois at Urbana-Champaign, ♯ The Ohio State University, ‡ Carnegie Mellon University ABSTRACT In this work, we address the challenge of encoding speech cap-tured by a microphone array Chao Weng, Dong Yu Tencent AI Lab, Bellevue, USA {cweng,dyu}@tencent. IEEE Signal Processing Magazine 2. research lab in Bellevue, Wash. J Yu, X Xie, S Liu, S Hu, MWY Lam, X Wu, KH Wong, X Liu, H Meng. The key component of this system is the Duration Informed Attention Network (DurIAN), an Wenhao Yu, Hongming Zhang, Xiaoman Pan, Peixin Cao, Kaixin Ma, Jian Li, Hongwei Wang, Dong Yu Tecent AI Lab wenhaowyu@global. krqvz xra awhc vxtt hvnvavj cipui onwn pqxw tblwy jlhz