A regular user of a multichannel loudspeaker system in typical living rooms sets the loudspeakers in a non-uniform manner, with angles and distances that don’t necessarily follow the recommended ITU-R BS.2159-4 standard. Assuming a multichannel audio system equipped with N number of loudspeakers and M very near-field (NF) microphones attached to each loudspeaker, the user location with respect to the loudspeakers can be estimated by utilizing a supervised machine learning (ML) model. Two neural networks (NN) were trained with the human speech directivity index (DI) computed by room simulations, where the sound source was the typical directivity radiation pattern of human speech, and the receivers were the NF microphones attached to the loudspeakers. The distances between loudspeakers and the DI data was combined as input for the two NN models. One network was dedicated to estimate distances from loudspeaker to user, and the other network was dedicated to the angle estimation. The results shown a 95% confidence interval (CI) of ±1.7 cm and a CI of ±7 degrees for the incidence angle.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=22292
(1254KB)
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.
