Blind Audio Source Separation (BASS) algorithms are often employed in applications where the aim is the acoustic reproduction of the separated source signals. The perceived quality of the reproduced signals is therefore a crucial criterion. Two different factors can roughly be distinguished which have influence on the perceived quality of blindly separated source signals. First, the quality of the separation of a desired target source from a signal mixture. Second, the preservation of the spatial image of the source, the spatial position of the target source in the signal mixture as it is perceived by the listener. Based on extensive MUSHRA-style listening tests, results are presented reflectling the influence of both factors on the overall basic audio quality of BASS signals. Further, a nonlinear regression model is set up to parametrize the influence of both factors on the subjective audio quality. A correlation of 0.98 between predicted and measured subjective quality and a root mean square prediction error of 2.7 on a [0,100] MUSHRA-scale was achieved for predicting the basic audio quality from an unknown listening test.
Author:
Kastner, Thorsten
Affiliations:
University of Erlangen-Nuremberg, Erlangen, Germany; Fraunhofer Institute for Integrated Circuits IIS, Erlangen, Germany(See document for exact affiliation information.)
AES Convention:
129 (November 2010)
Paper Number:
8299
Publication Date:
November 4, 2010
Subject:
Perception and Subjective Evaluation of Audio
Click to purchase paper as a non-member or you can login as an AES member to see more options.
No AES members have commented on this paper yet.
To be notified of new comments on this paper you can subscribe to this RSS feed. Forum users should login to see additional options.
If you are not yet an AES member and have something important to say about this paper then we urge you to join the AES today and make your voice heard. You can join online today by clicking here.