A computationally efficient model for combined assessment of monaural and binaural audio quality

Audio quality is an important aspect of hearing aids, hearables, and sound reproduction systems because the signal processing of such devices might alter the spectral composition or interaural differences of the original sound and thus might degrade the perceived audio quality.Consequently, an audio quality model applicable to such devices requires accounting for monaural and binaural aspects of audio quality.Flener et al. successfully predicted overall audio quality by combining a monaural and binaural audio quality model, which is computationally expensive and thus limits the scope of application.In order to also cover timecritical applications, such as real-time control of algorithms in audio and hearing technology, the authors present a computationally efficient model for overall audio quality in listeners with normal hearing.The suggested model was evaluated with six databases including quality ratings for music and speech signals processed by loudspeakers and algorithms typically applied in modern hearing devices (e.g., acoustic transparency, feedback cancellation or binaural beamforming).The presented model achieved a high prediction performance, indicated by the mean Pearson correlation of 0.9 similar to the more complex model of Flener et al., while its calculation time is substantially lower.