【阿里巴巴大模型产品“通义听悟”升级，引领音视频问答新纪元】今日，阿里推出其大模型产品“通义听悟”的全新升级，一口气上线了六大创新功能，其中包括备受瞩目的音视频问答助手“小悟”。这一升级标志着智能问答技术在超长音视频处理领域实现了重大突破。
“小悟”作为本次升级的核心亮点，凭借其多语言 Query 处理能力、长篇章文本理解技术、指令演化框架优化以及检索增强生成算法，开创性地支持了对超长音视频的单记录、跨记录、多语言自由问答。这意味着用户现在可以轻松地从长达数小时的音视频内容中，直接通过提问获取关键信息，无需再耗费大量时间进行手动检索和筛选。
此外，通义听悟的其他新功能也十分引人关注，如一键 AI 改写和思维导图生成等，进一步提升了内容处理的效率和质量。这些功能的推出，不仅在业界树立了新的标杆，也为用户提供了更为智能化和便捷的内容管理和分析工具。
据IT之家报道，此次“通义听悟”的升级，打破了音视频问答的行业上限，无论是支持的音视频时长还是文件数量，都创造了新的记录。这无疑将推动媒体、教育、研究等多个领域的工作效率，引领智能问答技术的新一轮革新。
英语如下：
**News Title:** “Tongyi Tingwu undergoes major upgrade: Introduces ultra-long video Q&A and unlocks AI mind map capability”
**Keywords:** Tongyi Tingwu upgrade, ultra-long video Q&A, Xiao Wu assistant
**News Content:** **Alibaba’s large language model product “Tongyi Tingwu” upgrades, pioneering a new era in audio-video Q&A.** Today, Alibaba unveiled an extensive upgrade to its large language model product “Tongyi Tingwu,” launching six innovative features, prominently featuring the audio-video Q&A assistant “Xiao Wu.” This upgrade signifies a significant breakthrough in intelligent Q&A technology for processing ultra-long audio and video content.
As the centerpiece of this upgrade, “Xiao Wu” stands out with its multi-language query processing, long-form text understanding, command evolution framework optimization, and retrieval-augmented generation algorithm. It pioneers support for single-record, cross-record, and multi-language free Q&A in ultra-long audio and videos. Users can now easily extract key information from hours-long audio-video content by posing questions directly, eliminating the need for time-consuming manual searches and screening.
Furthermore, other new features of Tongyi Tingwu, such as AI rewriting with a single click and mind map generation, have also garnered attention, enhancing content processing efficiency and quality. These advancements set new industry standards and provide users with more intelligent and convenient tools for content management and analysis.
According to IT Home, the upgrade of “Tongyi Tingwu” breaks the industry ceiling for audio-video Q&A, setting new records in both supported audio-video durations and file numbers. This is poised to boost efficiency across various sectors, including media, education, and research, driving a new wave of innovation in intelligent Q&A technology.
【来源】https://www.ithome.com/0/756/690.htm
Views: 1