8月17-21日，国际音频领域顶级会议INTERSPEECH在荷兰鹿特丹举办。荣耀联合上海交通大学完成的 两篇聚焦端侧多语种任务的研究成果成功入选INTERSPEECH2025录用论文，并受邀在会议上作技术发 表。作为全球语音科学与技术领域最具权威性的学术会议之一，INTERSPEECH的认可体现了荣耀在端 侧AI语音技术领域的持续努力与技术积累，这表明荣耀在全球化的AI技术交流中，取得了一定进展。 双论文入选国际顶会 INTERSPEECH作为国际音频领域顶级会议，其收录论文代表着全球语音技术研究的最前沿方向。荣耀 的两篇论文成功入选，聚焦的正是当前端侧AI语音技术的核心难题——如何在移动设备有限的算力与 存储资源下，实现媲美云端的多语种实时语音识别与通话翻译体验。 荣耀两位AI专家在荷兰鹿特丹INTERSPEECH学术交流现场作技术发表 荣耀在端侧AI语音技术领域的突破，并非偶然，而是品牌长期深耕AI战略的必然结果。自阿尔法战略 公布以来，荣耀在AI技术领域的投入始终保持"持续性"与"前瞻性"，从AI体验落地到技术开源，再到端 侧语音大模型突破，形成了清晰的战略演进路径。 此前，在世界人工智能大会（WAI ...

Group 1 - The 2025 World Artificial Intelligence Conference (WAIC) was held in Shanghai from July 26 to 28, featuring the launch of the "Global Artificial Intelligence Innovation Governance Center" [2] - The conference theme was "Intelligent Era, Global Cooperation," with five main components: forums, exhibitions, competitions, application experiences, and innovation incubation [2] - The exhibition area exceeded 70,000 square meters for the first time, attracting over 800 companies and showcasing more than 3,000 cutting-edge exhibits, marking the largest scale in the event's history [2] Group 2 - Major companies such as Alibaba, Tencent, Baidu, and iFlytek participated, with Alibaba unveiling its first self-developed AI glasses, "Quark AI Glasses," expected to launch within the year [3] - Baidu showcased three core AI technologies and products, including "Luo Bo Kuai Pao," "PaddlePaddle," and "Baidu Intelligent Computing Cluster," along with various other AI platforms and tools [3] - iFlytek presented its voice simultaneous interpretation model and various AI applications, demonstrating a response time of 2 seconds for Chinese-English translation [4] Group 3 - Shenzhen-based hard-tech companies also showcased their products, with Huawei presenting its Ascend 384 super node, highlighting its innovative capabilities and solutions across various industries [5] - Tencent displayed its "AI Full Family Bucket," featuring multiple first-time technologies and five major AI productivity platforms [5] - Honor launched its self-developed multi-modal perception model, "MagicGUI," and announced its open-source initiative for global developers [5]

Core Insights - The article emphasizes that the era of AI in smartphones should extend beyond basic functionalities like translation and document processing, advocating for a broader imagination of AI's capabilities in mobile devices [1] - Honor's release of the MagicGUI model, with 7 billion parameters, marks a significant advancement in AI assistants, evolving from traditional voice assistants to more capable digital assistants that can understand complex needs and execute multi-step tasks [1][2] Group 1: Evolution of AI Assistants - Since the rise of large models in 2023, major smartphone manufacturers have recognized the shift from basic voice assistants to lightweight intelligent agents capable of perception, reasoning, decision-making, and operation [2] - Honor's YOYO has evolved from executing single tasks to coordinating multiple intelligent agents, showcasing a significant leap in functionality compared to traditional voice assistants [2][7] Group 2: Comparison with Competitors - Apple's Siri, introduced in 2011, has seen limited updates and remains largely underutilized, while Android counterparts like Honor's YOYO, Vivo's "Blue Heart Little V," and Xiaomi's "Super Xiao Ai" have advanced to task-oriented intelligent agents capable of executing complex tasks [5][6] - The transition from app-driven interactions to agent-driven frameworks signifies a major shift in user-device interaction, with AI assistants taking the lead in understanding and executing tasks [8] Group 3: Technical Advancements - The MagicGUI model employs a two-phase training paradigm, enhancing the model's screen perception and positioning capabilities through large-scale GUI knowledge injection and reinforcement learning [9] - The trained MagicGUI model allows YOYO to think and act based on visual information from the screen, improving its efficiency and adaptability in task execution [9] Group 4: Open Source Initiative - Honor has announced that the MagicGUI model and related testing data will be made available on open-source platforms, promoting collaboration and further development in the field [9]

Core Viewpoint - Honor officially launched its self-developed multimodal perception model, MagicGUI, during the World Artificial Intelligence Conference (WAIC), marking a significant milestone in its Alpha strategy and aiming to enhance the AI ecosystem for global developers [1][10]. Group 1: Technology Innovation - MagicGUI features a parameter scale of 7 billion, achieving a 91.5% accuracy rate in common scenarios, which is a 16.4% improvement over top open-source models in the industry [2][3]. - The model employs a unique "continued pre-training + reinforcement fine-tuning" training scheme, addressing existing technical bottlenecks and enhancing data utilization efficiency and generalization capabilities [4][10]. Group 2: AI Application and User Experience - The Magic V5, equipped with the MagicGUI model, allows the YOYO assistant to autonomously manage tasks across applications, providing a seamless user experience [6][9]. - YOYO can execute complex tasks with a single command, such as booking rides through various apps, showcasing the model's multimodal perception and automation capabilities [9][10]. Group 3: Security and Compliance - Honor emphasizes user privacy and security, having obtained multiple international certifications, ensuring that AI operations are conducted with a focus on user data protection [9][11]. - The company is actively promoting the establishment of an AI safety governance system, collaborating with industry leaders to enhance transparency and practical implementation of AI safety measures [11]. Group 4: Open Collaboration and Ecosystem Development - Honor is committed to an open and collaborative AI ecosystem, sharing technical reports and core elements of the MagicGUI model to facilitate innovation and reduce barriers for global developers [12][14]. - The company has partnered with Fudan University to establish a joint laboratory for natural language processing, reflecting its belief in ecosystem collaboration for AI advancement [13][14].