A Study on the Audio Comparison Expletive Filtering System using TTS

Abstract

Objectives: The use of expletives on online is increasing more and more. As online is used by various age groups, preventing the use of expletives is required. The purpose of this study is to use TTS and wave audio files to increase the accuracy of the existing language filtering. Methods/Statistical Analysis: In the Default setting, the text of the expletive BASE data that becomes the filtering standard is stored on the DB using TTS that reads the text into audio and the visualization program that produces the audio file into a wave file. When the text is inputted by the user, the device reads the text through the TTS Program, and the speech is visualized to be produced into a wave file. The inputted text is searched in the base data. The produced wave file and the wave file of the base data are compared, and when the match rate is 80% or higher, it is inserted into the DB. Findings: The proposed system is for increasing the accuracy of the language filtering of the expletives which are used in the online chat of online games. There were various studies on the expletives, but for the filtering on swear words, the swear filter is mostly used. However, there are limits in detecting swear words that are modified. Therefore, it is expected that this system will enable the improvement of detecting modified swear words to lead to healthy use of language through online. Improvements/Applications: It is considered that further studies are required to improve the expletive search engine to strengthen the restriction of expletive use, and to collect accurate and various expressions of expletive data to enable real-time, automated language filtering.