An adaptive approach to the analysis and synthesis of musical signals is presented. These musical signals are modeled as nonstationary time-varying signals with mixed spectra. A frame-by-frame-based analysis and synthesis approach is adopted to capture the important time-varying nature of the musical signals. Through data analysis the relationship between a number of factors is discussed, such as modeling accuracy and efficiency, frame length, model order, and accuracy of parameter estimation. Once the signal parameters have been extracted from data analysis, the sounds are synthesized using these parameters. Analysis and synthesis results using variable model orders and variable frame lengths in combination with high-resolution parameter estimation and residual modeling are presented. The results show that the proposed algorithms are computationally efficient and effective when synthesizing piano and trumpet tones.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.