Team MTS @ AutoMin 2021: An Overview of Existing Summarization Approaches and Comparison to Unsupervised Summarization Techniques

Description of the datasets used and of the designed approaches; Remote communication through video or audio conferences has become more popular than ever because of the worldwide pandemic. Experiment procedure and results; These events, therefore, have provoked the development Conclusion and further work. of systems for automatic minuting of spoken language leading to AutoMin 2021 challenge. The following paper illustrates the In this project we aim to achieve three main goals. First, we results of the research that team MTS has carried out while participating wish to analyze the existing pre-trained summarization models in the Automatic Minutes challenge. In particular, in and compare their performance in summarization of manually this paper we analyze existing approaches to text and speech transcribed audio recordings. Second, we propose a custom unsupervised summarization, propose an unsupervised summarization technique approach for summarization of English texts. Last based on clustering and provide a pipeline that includes but not least, we adapt our summarization module for multichannel an adapted automatic speech recognition block able to run on meeting audio recordings and made the pipeline opensource.