Voice data heat analysis method and system

Abstract

The invention discloses a voice data thermal analysis method and system. The method comprises: receiving voice data; performing endpoint detection on the voice data to obtain valid voice segments and non-effective voice segments in the voice data; Perform speech recognition on the effective speech segments to obtain the recognition text corresponding to the effective speech segments; extract the heat characteristics based on acoustics for each effective speech segment; extract each effective speech segment based on the content according to each effective speech segment and its corresponding recognition text The heat feature on the above; using the heat feature based on the acoustics and the heat feature based on the content, and the pre-built heat prediction model, calculate the first heat value of each effective speech segment. The invention can improve the accuracy of heat analysis of voice data.

Description

technical field [0001] The invention relates to the field of voice signal processing, in particular to a voice data thermal analysis method and system. Background technique [0002] Heat, as the name suggests, refers to the degree of heat. In the field of information retrieval, heat is usually used to represent the degree of people's interest in the search object, and heat analysis can help users quickly find the content they are interested in. For text data, you can generally use the keywords it contains to determine its popularity; and for voice data, if it is converted into text through speech recognition, and then determine the popularity of the corresponding voice data according to the keywords in the text, due to the increase Without the process of speech recognition, speech recognition will not only produce recognition errors, but also lose the audio information in the speech data in the recognized text, which will lead to the problem of reducing the accuracy of heat...

AI Technical Summary

Problems solved by technology

For example, for conference recording data, if the user wants to know which segment of voice data is the most intense part of the conference discussion and which segment of voice data is hardly spoken, it is necessary to play back all the voice data to determine. When the voice data is long, it will obviously give It brings great troubles to users, unable to quickly find the needed voice data segment, and the efficiency is low

Images