Article Details

Abstract

With the popularization of the hybrid office model, the pain points of traditional conference systems in terms of transcription efficiency, minutes quality and data management have become increasingly prominent, which are difficult to meet the needs of intelligent office. To this end, this paper carries out the research on the design and implementation of an intelligent conference system. Firstly, by analyzing the core needs of conference transcription, minutes generation and data management, a five-layer system architecture including the infrastructure layer, technology layer, data layer, application layer and interaction layer is designed, and the functions and collaborative relationships of each layer are clarified. Secondly, based on this architecture, the design of three core functions is completed: conference transcription, conference minutes generation and conference data management. In terms of key technologies, voice recognition engine, voiceprint extraction technology, large language model and vector database are integrated to optimize the system performance and intelligence level. Finally, the system development and deployment are completed based on the front-end and back-end separation and microservice architecture, and through functional verification experiments, it is proved that each module of the system runs stably, which can effectively improve the conference processing efficiency and knowledge precipitation capacity, providing a feasible technical solution for intelligent conferences in the hybrid office scenario.

Keywords

intelligent conference system; speech recognition; large language model; semantic retrieval

References

1. [1] Shi Xiaohu, Yuan Yuping, Lv Guilin, etc Overview of Compression Algorithms for Automatic Speech Recognition Models [J]. Journal of Jilin University (Science Edition), 2024, 62 (1): 122-131
2. [2] His Rilige, Siqintu, Wangsriguleng A review of research on automatic text summarization based on deep learning [J]. Computer Engineering and Applications, 2025, 61 (18): 24-40
3. [3] Zeng Chunyan, Ma Chaofeng, Wang Zhifeng, etc A review of speaker recognition research under deep learning framework [J]. Computer Engineering and Applications, 2020, 56 (7): 8-16
4. [4] Bao Changchun, Yang Xue Research Status and Prospects of Multi Speaker Separation and Target Speaker Extraction [J]. Data Collection and Processing, 2024, 39 (5): 1044-1061
5. [5] Huang Feifei Research on BERT's Graph Model Text Abstract Generation Method [J]. Modern Information Technology, 2022, 6 (2): 91-95100
6. [6] He Junmin, Lu Menghua, Meng Kui Chinese Single Document Abstraction Model DSum SSE [J]. Computer Engineering and Applications, 2021, 57 (15): 200-206
7. [7] Xiao Lei, Chen Zhen's family A review of data-driven Chinese entity extraction methods [J]. Computer Engineering and Applications, 2024, 60 (16): 34-48
8. [8] Wang Jingyu, Luan Junqing, Tan Yuesheng Research on Big Data Access Control Model Based on Data Sensitivity [J]. Computer Engineering and Applications, 2019, 55 (23): 70-77
9. [9] Tong Qian, He Heng, Nie Lei, etc Hierarchical Lightweight Access Control Scheme in Cloud Environment [J]. Computer Engineering and Applications, 2022, 58 (21): 109-118
10. [10] Zhang Xiaodong, Chen Taowei, Yu Yimin Distributed Data Access Control and Sharing Scheme Based on CPABE [J]. Computer Engineering and Applications, 2023, 59 (23): 254-262
11. [11] Wu Xu, Bian Wenqiang, Jie Xiaqing, etc Machine Reading Comprehension based Chinese Event Extraction Method [J]. Computer Engineering and Applications, 2023, 59 (16): 93-100
12. [12] Wang Jinrui, Li Jie Chinese event extraction method integrating event types [J]. Computer Engineering and Applications, 2025, 61 (1): 214-220
13. [13] Liu Xueying, Yun Jing, Li Bo, etc A review of retrieval enhanced generation based on large-scale language models [J]. Computer Engineering and Applications, 2025, 61 (13): 1-25
14. [14] Zhou Chao, Shi Xiaosu, Yang Fan, etc Research on Intelligent Classification and Self Dispatch Methods Based on Semantic Analysis [J]. Modern Information Technology, 2025, 9 (20): 44-50, 54

Funding

China Railway Information Technology Group Co., Ltd. Technology Research and Development Plan WJZG-CKY-2024012 (2024K09)

About the authors

Wang Ziteng (1993-), female, Han ethnicity, from Beijing, senior engineer, master's student, research interests include railway informatization and artificial intelligence.