Sign language translation system and method for converting voice of video into avatar and animation

摘要

The present invention relates to a sign language translation system and method for converting the audio of an image into an avatar and animation. By enabling the output of the avatar and animation generated by the output unit, the user must directly input input data to translate into sign language. By solving the problem, the problem of translating into sign language was solved when the existing user had to directly input the input data. That is, in the present invention, in a sign language translation program, a voice extracting unit constructed using No. 201 to extract audio from an image, a text conversion unit configured by extracting features from the extracted voice so that the extracted audio can be converted into text, The converted text is converted into a text refiner that is composed of morphemes by purifying the converted text to output accurate information, and the refined text to be translated into a sign language using deep learning technology to translate the refined text into a sign language. The configured sign language translation unit, the sign language animation conversion unit configured by converting the sign language into animation by learning the data about the sign language to the model so that the translated sign language can be generated as an avatar and animation, and the generated avatar and animation can be output. An output section consisting of placing the converted animation at the bottom of the screen, a signal analysis section consisting of extracting the audio of a video to analyze the audio signal, and a voice consisting of the analyzed audio with accurate data so that the analyzed audio can be refined. A feature extraction unit consisting of extracting the features of the voice that is related to the voice data so that the features of the analyzed voice can be extracted, and a score calculation of the features extracted from the voice data so that the SCORE for the voice features can be calculated. The configured SCORE extraction unit, the SCORE selection unit configured by selecting the score with the maximum value from the result of calculating the score so that the feature with the maximum SCORE can be selected, calculates the score so that the selected features can be combined, and has the maximum score. It is composed of a post-processing unit configured by combining elements, and an output unit configured by combining elements and outputting the combined result on the screen so that the combined result can be output. Accordingly, the present invention solves the problem of translating into sign language only when the user directly inputs input data by enabling the output of the avatar and animation generated by the output unit, so that the existing user must directly enter the input data to translate into sign language. It has the effect of solving the problem of being