Audio Summarization with Audio Features and Probability Distribution Divergence

Abstract

The automatic summarization of multimedia sources is an important task that facilitates the understanding of an individual by condensing the source while maintaining relevant information. In this paper we focus on audio summarization based on audio features and the probability of distribution divergence. Our method, based on an extractive summarization approach, aims to select the most relevant segments until a time threshold is reached. It takes into account the segment's length, position and informativeness value. Informativeness of each segment is obtained by mapping a set of audio features issued from its Melfrequency Cepstral Coefficients and their corresponding Jensen-Shannon divergence score. Results over a multi-evaluator scheme shows that our approach provides understandable and informative summaries.

A Comparative Study for Speech Summarization Based on Machine Learning: A Survey

“…It is based on employing textual information to learn an in formativeness representation based on probability distribution divergences, which is not taken into account by normal audio summarization with audio features. The length of the summary was set to equal 35% of the original audio length [16]. Weng, S.-Y., Lo, T.-H., Chen, B. extended the BERT-based method for supervised extractive speech summarization that is capable of performing robust summary on spoken documents containing erroneous ASR transcripts.…”

“…1A subjective scaled opinion metric of 1-5 was used to assess the quality of the generated summaries and their components. Two objective metrics were also used: full score and average score metrics [16]. In order to obtain an overall score suitable for DEDA method, ROUGE scores for every meeting transcript in the test set were computed and then the macro-averaging method was used in [20].…”