Nexdata.ai

🎥 Keynote Replay | 2nd MLC-SLM Workshop Speech is more than just words. How can AI models move beyond transcription to better understand long-form speech and conversations? At the 2nd MLC-SLM Workshop, held alongside INTERSPEECH 2026, Dr. Jinyu Li, Partner Applied Science Manager at Microsoft, IEEE Fellow, and 2026 ISCA Fellow, delivered an insightful keynote: 🎤 "Long-form Speech Modeling: Beyond Transcription" As Speech AI continues to advance, modeling long-form speech presents new challenges and opportunities beyond traditional speech recognition. Dr. Li's talk explores the evolving direction of speech modeling and the importance of capturing richer information from spoken language. 📺 The full keynote is now available on YouTube! Watch the replay and explore what's next for Speech AI. 🔗 youtu.be/iW9oJLd6M78 #INTERSPEECH2026 #MLCSLM #SpeechAI #SpeechLLM #LongFormSpeech #SpeechRecognition #ConversationalAI #AIResearch #Nexdata

🎥 Keynote Replay | 2nd MLC-SLM Workshop How can we make foundation models better aligned with human needs, preferences, and expectations? At the 2nd MLC-SLM Workshop, held alongside INTERSPEECH 2026, Prof. Shinnosuke Takamichi, Associate Professor at Keio University, Japan, delivered an insightful keynote: 🎤 "Aligning Foundation Models with Humans" As foundation models continue to advance, aligning AI systems with human expectations is becoming increasingly important. Drawing on his expertise in speech and audio processing, Prof. Takamichi explored the relationship between foundation models and humans, offering valuable perspectives on the future of human-centered AI. 📺 The full keynote is now available on YouTube! Watch the replay and revisit the discussion. 🔗 youtu.be/wCULh0Hh7FI #INTERSPEECH2026 #MLCSLM #SpeechAI #FoundationModels #HumanCenteredAI #SpeechResearch #GenerativeAI #AIResearch #Nexdata

🎥 Keynote Replay | 2nd MLC-SLM Workshop Are higher benchmark scores always a sign of better AI models? What can model evaluations tell us beyond the numbers? At the 2nd MLC-SLM Workshop, held alongside INTERSPEECH 2026, Prof. Hung-yi Lee, Professor at National Taiwan University (NTU) and ISCA Fellow, delivered an insightful keynote: 🎤 "Beyond the Benchmark Score: Lessons from Evaluating Models" As AI models continue to evolve, evaluating their capabilities goes far beyond comparing benchmark scores. Prof. Lee's talk explores the lessons learned from model evaluation and raises important questions about how we measure AI progress. 📺 Missed the session? Watch the full keynote on YouTube and revisit the discussion! 🔗youtu.be/0sAFZKBXXQ8 #INTERSPEECH2026 #MLCSLM #SpeechAI #ModelEvaluation #AIResearch #Benchmarking #LLM #Nexdata

Five synchronized devices. 10,000 hours of human manipulation data. Nexdata recently completed an Ego-Exo data collection project for a leading AI company, capturing real tasks across factories, warehouses, and other working environments for robot learning. Each setup combined a stereo head-mounted device, two wrist cameras, and two external cameras—connecting close-up hand–object interactions with full-body movements. From scene diversity to synchronization and quality checks, what does collection at this scale involve? Read the case study 👇 nexdata.ai/company/news/1478… #Nexdata #PhysicalAI #RobotLearning #Robotics #EgocentricData #EgoExo #HumanDemonstration #DataCollection #AITrainingData

How can continuous autoregressive modeling advance speech generation? At the 2nd MLC-SLM Workshop, Professor Kai Yu of Shanghai Jiao Tong University—ISCA Fellow (2025) and IEEE Fellow (2026)—shared his insights in “Continuous Autoregressive Foundation Model for Speech Generation.” Watch the full presentation on YouTube to explore the approach and its potential for next-generation speech models. ▶️ Watch here: youtube.com/watch?v=Jo3pE8hN… #SpeechAI #SpeechGeneration #AutoregressiveModels #FoundationModels #SpeechTechnology #ISCA #MLCSLM #Nexdata

Building Speech AI for different languages and real-world conversations starts with the right training data. In her presentation at the 2nd MLC-SLM Workshop, Nexdata Sales Manager You Wu introduces our Speech AI Data Solution, covering multilingual speech datasets, Bespoke data collection, and annotation services. The full talk is now on YouTube. Explore how we support the data needs of advanced Speech AI models. ▶️ Watch here: youtu.be/qdxiOqYndQc #SpeechAI #MultilingualAI #SpeechData #TrainingData #DataCollection #DataAnnotation #MLCSLM #Nexdata

Building a Physical AI data pipeline means getting hardware, software, and data operations to work together. Nexdata brings collection devices, data management and annotation platforms, and Bespoke services into one integrated solution—available as individual components or a turnkey deployment. 🔗 Click link to explore how we help teams capture, manage, and scale their Physical AI training data. nexdata.ai/company/news/1476… #PhysicalAI #Robotics #RobotLearning #EmbodiedIntelligence #DataCollection #DataAnnotation #TrainingData #Nexdata

🎥 Keynote Replay | 2nd MLC-SLM Workshop In the era of Large Language Models (LLMs), what compelling research challenges remain for the speech community? At the 2nd MLC-SLM Workshop, held alongside INTERSPEECH 2026, Prof. Yonghong Yan, Chief Scientist at the Institute of Acoustics, Chinese Academy of Sciences, delivered a thought-provoking keynote: 🎤 "In the LLM Era: Do Speech Researchers Still Have Compelling R&D Topics?" With decades of experience in speech signal processing and recognition, Prof. Yan shared his perspectives on the evolving landscape of speech research and the opportunities that remain in the age of LLMs. 📺 The full keynote is now available on YouTube! Watch the replay and explore the future of speech research. Link here: youtu.be/ys2yr75L_Co #MLCSLM #INTERSPEECH2026 #SpeechAI #LLM #SpeechResearch #Nexdata

🎥 Missed Nexdata's presentation at IROS 2026? The full talk is now available on YouTube! At IROS 2026 in Pittsburgh, Nexdata presented our Physical AI Data Solution, showcasing how an integrated, end-to-end data pipeline can accelerate robotics development. From proprietary data collection devices and large-scale data collection to data management, annotation, and model-ready datasets, Nexdata provides a one-stop Physical AI Data Solution designed to streamline the entire workflow. Our presentation highlights: 🔹 Teleoperation, Ego-Centric & UMI Data Collection 🔹 Simulation Data & Scan-to-Sim 🔹 Proprietary Data Collection Devices 🔹 Integrated Data Management & Annotation Platforms Whether you're building VLA models or advancing robotic manipulation, discover how Nexdata supports scalable, high-quality Physical AI data production. ▶️ Watch the full presentation on YouTube: youtu.be/VDSaC6MpS7M #Nexdata #IROS2026 #PhysicalAI #Robotics #RobotLearning #AIData

INTERSPEECH 2026 has wrapped up — but the conversation doesn’t end here. On October 2, the 2nd MLC-SLM Workshop will continue the discussion at ICC Sydney, bringing together leading researchers to share new perspectives on speech modeling, generation, evaluation, and foundation models. This year’s keynote lineup includes: 🎤 𝗬𝗼𝗻𝗴𝗵𝗼𝗻𝗴 𝗬𝗮𝗻 𝘞𝘩𝘢𝘵 𝘌𝘭𝘴𝘦 𝘊𝘢𝘯 𝘞𝘦 𝘋𝘰 𝘪𝘯 𝘵𝘩𝘦 𝘌𝘳𝘢 𝘰𝘧 𝘓𝘢𝘳𝘨𝘦 𝘔𝘰𝘥𝘦𝘭𝘴 🎤 𝗝𝗶𝗻𝘆𝘂 𝗟𝗶 𝘓𝘰𝘯𝘨-𝘧𝘰𝘳𝘮 𝘚𝘱𝘦𝘦𝘤𝘩 𝘔𝘰𝘥𝘦𝘭𝘪𝘯𝘨: 𝘉𝘦𝘺𝘰𝘯𝘥 𝘛𝘳𝘢𝘯𝘴𝘤𝘳𝘪𝘱𝘵𝘪𝘰𝘯 🎤 𝗞𝗮𝗶 𝗬𝘂 𝘊𝘰𝘯𝘵𝘪𝘯𝘶𝘰𝘶𝘴 𝘈𝘶𝘵𝘰𝘳𝘦𝘨𝘳𝘦𝘴𝘴𝘪𝘷𝘦 𝘍𝘰𝘶𝘯𝘥𝘢𝘵𝘪𝘰𝘯 𝘔𝘰𝘥𝘦𝘭 𝘧𝘰𝘳 𝘚𝘱𝘦𝘦𝘤𝘩 𝘎𝘦𝘯𝘦𝘳𝘢𝘵𝘪𝘰𝘯 🎤 𝗛𝘂𝗻𝗴-𝘆𝗶 𝗟𝗲𝗲 𝘉𝘦𝘺𝘰𝘯𝘥 𝘵𝘩𝘦 𝘉𝘦𝘯𝘤𝘩𝘮𝘢𝘳𝘬 𝘚𝘤𝘰𝘳𝘦: 𝘓𝘦𝘴𝘴𝘰𝘯𝘴 𝘧𝘳𝘰𝘮 𝘌𝘷𝘢𝘭𝘶𝘢𝘵𝘪𝘯𝘨 𝘔𝘰𝘥𝘦𝘭𝘴 🎤 𝗦𝗵𝗶𝗻𝗻𝗼𝘀𝘂𝗸𝗲 𝗧𝗮𝗸𝗮𝗺𝗶𝗰𝗵𝗶 𝐴𝑙𝑖𝑔𝑛𝑖𝑛𝑔 𝐹𝑜𝑢𝑛𝑑𝑎𝑡𝑖𝑜𝑛 𝑀𝑜𝑑𝑒𝑙𝑠 𝑤𝑖𝑡𝒉 𝐻𝑢𝑚𝑎𝑛𝑠 Beyond the keynote sessions, the workshop will also feature the 2nd MLC-SLM Challenge Summary & Awards, oral presentations, poster sessions, and more discussions around multilingual conversational speech and Speech LLMs. 📅 October 2, 2026 📍 ICC Sydney | C2.2 & C2.3 Official registration has closed, but on-site registration is still available. Still in Sydney after INTERSPEECH? Join us for another full day of research, ideas, and conversations around the future of speech and language AI. #INTERSPEECH2026 #MLCSLM #SpeechAI #SpeechLL #ConversationalAI #MultilingualAI #FoundationModels #Nexdata

Day 1 at IROS 2026 has wrapped up! 🤖 It was a busy first day in Pittsburgh, with great conversations around Physical AI, robot learning, and Ego-Centric data. At the Nexdata booth, our multi-scenario Ego-Centric data drew strong interest from researchers and industry experts. We had many discussions around how Ego data can capture human interactions and task execution across diverse environments, including homes, factories, offices, retail stores, warehouses, and more. A great start to IROS 2026 — and there’s much more to come. 📍 Nexdata | Booth #640 📅 IROS 2026 | Pittsburgh Stop by our booth to explore our latest Physical AI data solutions and see the Ego-Centric data capture device in action. #IROS2026 #Nexdata #PhysicalAI #EgoCentric #Robotics #RobotLearning #AIData

Day 1 at INTERSPEECH 2026 has wrapped up! 🎙️ It was a great start in Sydney, with lively conversations throughout the day around multilingual speech, conversational AI, Speech LLMs, and the future of human-machine interaction. At the Nexdata booth, we were glad to meet researchers, professors, and industry experts from around the world. Multi-speaker conversations and multilingual speech were among the topics that sparked particularly active discussions, with many interesting perspectives shared on data, models, and emerging research directions. And this is just the beginning. 📍 Nexdata | Booth #E5 📅 INTERSPEECH 2026 | Sydney If you’re attending INTERSPEECH, stop by Booth #E5 and continue the conversation with us! #INTERSPEECH2026 #Nexdata #ConversationalAI #SpeechAI #SpeechLLM #MultilingualAI

A special moment at 𝗜𝗡𝗧𝗘𝗥𝗦𝗣𝗘𝗘𝗖𝗛 𝟮𝟬𝟮𝟲 today — congratulations to 𝗝𝗶𝗻𝘆𝘂 𝗟𝗶 on receiving his 𝟮𝟬𝟮𝟲 𝗜𝗦𝗖𝗔 𝗙𝗲𝗹𝗹𝗼𝘄 𝗿𝗲𝗰𝗼𝗴𝗻𝗶𝘁𝗶𝗼𝗻 at today’s Fellow Ceremony! 👏 isca-speech.org/Latest-News/… We’re especially excited to celebrate this moment as Dr. Li will also be joining us as a 𝗸𝗲𝘆𝗻𝗼𝘁𝗲 𝘀𝗽𝗲𝗮𝗸𝗲𝗿 𝗮𝘁 𝘁𝗵𝗲 𝟮𝗻𝗱 𝗠𝗟𝗖-𝗦𝗟𝗠 𝗪𝗼𝗿𝗸𝘀𝗵𝗼𝗽 𝗼𝗻 𝗢𝗰𝘁𝗼𝗯𝗲𝗿 𝟮. This year, we are honored to welcome 𝘁𝘄𝗼 𝗜𝗦𝗖𝗔 𝗙𝗲𝗹𝗹𝗼𝘄𝘀 to the workshop: 𝗝𝗶𝗻𝘆𝘂 𝗟𝗶 (𝟮𝟬𝟮𝟲) and 𝗞𝗮𝗶 𝗬𝘂 (𝟮𝟬𝟮𝟱). Together with Yonghong Yan, Hung-yi Lee, and Shinnosuke Takamichi, they will share their latest research and perspectives on multilingual conversational speech, Speech LLMs, and spoken language technologies. 📅 𝗢𝗰𝘁𝗼𝗯𝗲𝗿 𝟮, 𝟮𝟬𝟮𝟲 📍 𝗜𝗖𝗖 𝗦𝘆𝗱𝗻𝗲𝘆, 𝗖𝟮.𝟮 & 𝗖𝟮.𝟯 Still want to join us? Online registration through the INTERSPEECH registration system has now closed, but 𝗼𝗻-𝘀𝗶𝘁𝗲 𝗿𝗲𝗴𝗶𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 𝗶𝘀 𝘀𝘁𝗶𝗹𝗹 𝗮𝘃𝗮𝗶𝗹𝗮𝗯𝗹𝗲. If you’re attending INTERSPEECH in Sydney and would like to join the workshop, you can register with us on site. forms.cloud.microsoft/r/zh0Z… We look forward to seeing you at the 2nd MLC-SLM Workshop! #INTERSPEECH2026 #MLCSLM #ISCAFellow #SpeechLLM #ConversationalAI

⏳ 2 DAYS TO GO — IROS 2026! We’re almost ready for Pittsburgh! Come meet Nexdata at Booth #640 and discover our latest Physical AI data solutions. 🎤 Don’t miss our presentation: Physical AI Data Solution 📅 Tuesday, September 29 🕒 3:45–4:00 PM We’ll be sharing how Nexdata supports Physical AI development with Ego-Centric, UMI, Teleoperation, Simulation, and more. 🎁 And yes, we’ve got gifts! Stop by Booth #640 for Nexdata giveaways, including our limited Shine Bags, available each day starting at: 🕙 10:00 AM 🕑 2:00 PM Come early to grab yours! 📍 IROS 2026 | Booth #640 #IROS2026 #PhysicalAI #Robotics #RobotLearning #Nexdata

⏳ Countdown to IROS 2026! We’re getting ready for IROS 2026 in Pittsburgh, where Nexdata will showcase our latest Physical AI Data Solution for robot learning and model development. 🎤 Don’t miss our presentation: Physical AI Data Solution 📅 Tuesday, September 29 🕒 3:45–4:00 PM More Details: 2026.ieee-iros.org/program/i… We’ll introduce Nexdata’s latest Physical AI data capabilities, covering Teleoperation, UMI, Ego-Centric, Simulation, and scalable Bespoke data collection for diverse robotic tasks and environments. 📍 IROS 2026 | Pittsburgh 📅 September 27 – October 1 📌 Booth #640 See you at IROS 2026! #IROS2026 #PhysicalAI #Robotics #RobotLearning #AIData #Nexdata

🎤 Meet Our Keynote 4 Speaker: Hung-yi Lee We’re pleased to welcome Hung-yi Lee, Professor in the Department of Electrical Engineering at National Taiwan University, to the 2nd MLC-SLM Workshop, a satellite event of INTERSPEECH 2026. Prof. Lee’s research focuses on reducing the need for annotated data in speech processing and natural language processing, including voice conversion, speech recognition, summarization, and question answering. He also shares deep learning knowledge through his Mandarin-language YouTube channel, which has more than 300,000 subscribers. 📅 October 2, 2026 📍 ICC Sydney, Rooms C2.2 & C2.3 🎟️ Registration: AUD 80, including coffee breaks and one lunch. There are two ways to register: Via INTERSPEECH: Select “Workshop on 2nd Multilingual Conversational Speech Language Model” during official registration. Via the on-site channel: Complete the registration form and pay in cash at the venue. Learn more about our speaker and the Workshop program on the official website. #MLCSLM #INTERSPEECH2026 #SpeechAI #Nexdata

🎤 Meet Our Keynote 3 Speaker: Kai Yu We’re pleased to welcome Kai Yu, Distinguished Professor at Shanghai Jiao Tong University and Co-founder and Chief Scientist of AISpeech, to the 2nd MLC-SLM Workshop, a satellite event of INTERSPEECH 2026. Prof. Yu directs the university’s Institute of Machine Intelligence, with research focused on cognitive conversational agents for complex scenarios. An IEEE Fellow and ISCA Fellow, he has published over 300 peer-reviewed papers, with contributions spanning speech recognition, spoken dialogue systems, and the deployment of spoken language technology. 📅 October 2, 2026 📍 ICC Sydney, Rooms C2.2 & C2.3 🎟️ Registration: AUD 80, including coffee breaks and one lunch. There are two ways to register: Via INTERSPEECH: Select “Workshop on 2nd Multilingual Conversational Speech Language Model” during official registration. Via the on-site channel: Complete the registration form and pay in cash at the venue. Learn more about our speaker and the Workshop program on the official website. #MLCSLM #INTERSPEECH2026 #SpeechAI #Nexdata

🎤 Meet Our Keynote 2 Speaker: Jinyu Li We’re pleased to welcome Jinyu Li, Partner Applied Science Manager at Microsoft, to the 2nd MLC-SLM Workshop, a satellite event of INTERSPEECH 2026. Dr. Li leads a science team advancing speech, translation, and language technologies. His research includes Speech LLMs, end-to-end speech recognition, and speech translation. An IEEE Fellow and ISCA Fellow, he received the IEEE Signal Processing Society Best Paper Award in 2025. 📅 October 2, 2026 📍 ICC Sydney, Rooms C2.2 & C2.3 Registration: AUD 80, including coffee breaks and one lunch. There are two ways to register: Via INTERSPEECH: Select “Workshop on 2nd Multilingual Conversational Speech Language Model” during official registration：interspeech2026.org/en-AU/pa… Via the on-site channel: Complete the registration form and pay in cash at the venue：forms.cloud.microsoft/r/zh0Z… Learn more about our speaker and the Workshop program on the official website： nexdata.ai/competition/mlc-s… #MLCSLM #INTERSPEECH2026 #SpeechAI #Nexdata

🎤 Meet Our Keynote 1 Speaker: Yonghong Yan We’re pleased to welcome Yonghong Yan, Chief Scientist at the Institute of Acoustics, Chinese Academy of Sciences, to the 2nd MLC-SLM Workshop, a satellite event of INTERSPEECH 2026. With 40 years of experience in speech signal processing and speech recognition, he has published over 300 papers and holds more than 100 invention patents. His research has helped Baidu, Tencent, and Alibaba build their first speech applications. 📅 October 2, 2026 📍 ICC Sydney, Rooms C2.2 & C2.3 Registration: AUD 80, including coffee breaks and one lunch. There are two ways to register: • Via INTERSPEECH: Select “Workshop on 2nd Multilingual Conversational Speech Language Model” during official registration: interspeech2026.org/en-AU/pa… • Via the on-site channel: Complete the registration form and pay in cash at the venue: forms.cloud.microsoft/r/zh0Z… Learn more about our speaker and the Workshop program on the official website: nexdata.ai/competition/mlc-s… #MLCSLM #INTERSPEECH2026 #SpeechAI #Nexdata

🤖 Meet Nexdata at IROS 2026 in Pittsburgh! We’re excited to join IROS 2026, bringing Nexdata’s latest data resources and services for Physical AI to the global robotics community. At Booth #640, we’ll showcase different types of data supporting robot learning and model development, including Teleoperation, UMI, Ego-Centric, Simulation, World Model, and more. We’ll also introduce our Bespoke capabilities across data collection, data annotation, and model validation, supporting different data requirements throughout Physical AI model development. 📍 Pittsburgh, Pennsylvania, USA 📅 September 27 – October 1, 2026 📌 Booth #640 Heading to IROS 2026? Stop by Booth #640 and talk with our team about your Physical AI data needs. Better Data. Better Robots. nexdata.ai/company/news/1448… #IROS2026 #Nexdata #PhysicalAI #Robotics #RobotLearning #EgoCentric #Teleoperation #AIData