The present invention provides a multimodal model generating system for generating a learning model for detecting cough by combining multiple data, comprising: an input unit for receiving first data and second data having a two-dimensional structure; a conversion unit for generating converted data by transposing any one of the first data and the second data; And a model generator for generating a multi-modal learning model by thermally combining unconverted data among the first data or the second data with the converted data; to detect a cough when using machine learning or deep learning, including It is characterized by improving performance.
展开▼