This paper compares the conventional logarithmic sweep and a perfect sequence sweep (PSEQ) for simultaneous deconvolution of multichannel loudspeaker-room impulse responses using a single time domain measurement. The optimal parameters (including sweep duration) for each sweep are determined using machine learning (ML) techniques on training data of noise-free room impulse responses. Each of these optimized sweeps is evaluated on a noise-less and noisy room-response test set to assess simultaneous deconvolution performance in the absence/presence of noise. The results show that the optimized conventional log-sweep, while shorter in duration, is better suited for noisy conditions. In contrast, the PSEQ sweep works well under conditions with little background noise. The paper compares both optimized sweeps with the shortest-length bandwise minimum-noise (BMN) sweep. Both optimized sweeps (viz., conventional and PSEQ) produce lengths shorter than the BMN sweep for simultaneous deconvolution with a single time-domain measurement. Thus, depending on the measurement time requirement and background noise condition, on the basis of this research, there is a clear consensus on which sweep would be best suited for simultaneous deconvolution.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=22497
(11160KB)
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.
