This project explores granular synthesis techniques that utilize various basis functions inspired by existing matching pursuit algorithms. The first algorithm performs a STFT on an input signal and synthesizes a new, granular signal using one-dimensional Gabor atoms. These atoms can be made to virtually reproduce the input signal, but a wide variety of granular effects can be achieved by altering the distribution of the atoms in the time and frequency domains, such as granular time stretching and pitch shifting, along with statistical distribution techniques introduced by the author. The second algorithm utilizes a basis set of generated noise bursts, which can be over-complete or an orthonormal basis for the Hilbert space that corresponds to the analysis window by applying the Gram-Schmidt process to the burst library. The noise functions are then used as grain contents in the synthesis stage, where a variety of effects are created with redistribution methods. Audio examples are provided over headphones.
(See document for exact affiliation information.)
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 Join the AES. If you need to check your member status, login to the Member Portal.
