Adaptive Digital Audio Effects are sound transformations controlled by features extracted from the sound itself. Artificial reverberation is used by sound engineers in the mixing process for a variety of technical and artistic reasons, including to give the perception that it was captured in a closed space. We propose a design of an adaptive digital audio effect for artificial reverberation that allows it to learn from the user in a supervised way. We perform feature selection and dimensionality reduction on features extracted from our training data set. Then a user provides examples of reverberation parameters for the training data. Finally, we train a set of classifiers and compare them using 10-fold cross validation to compare classification success ratios and mean squared errors. Tracks from the Open Multitrack Testbed are used in order to train and test our models.
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.