Sound Source Decomposition and Modeling for Customer Product Expectation

This paper presents a systematic approach to decompose complex sounds, model individual sound components of the overall sound, and modify the resulting sound component model to tune for optimal customer satisfaction. A technique has been developed which performs Dziewonski time-frequency analysis of a complex sound. The complex sound structure is visually decomposed into individual sound components. With the sound decomposed, individual sound components are extracted for further analysis. A model template is chosen for a particular sound structure. The parameters of the model are optimally fitted to the extracted sound time, frequency and amplitude characteristics. A suite of time series can be generated by varying parameters of the sound model. Audio playback of the generated time series lends to subjective analysis such as paired comparison for optimal customer satisfaction.