Audio Segmentation Method based on Attention Mechanism

摘要

An audio segmentation method based on an attention mechanism is provided. In the audio segmentation method according to an embodiment of the present invention, a mapping relationship between an audio spectrum feature vector for generating an audio signal that is automatically synthesized using the input text and the input text is obtained, and input using the mapping relationship Split the audio signal. Accordingly, through the audio division using the attention mechanism, it is possible to significantly reduce effort, time, and cost while being high quality.