1. Subjective Assessments of Real-Time Room Dereverberation and Loudspeaker Equalisation Panagiotis Hatziantoniou and John Mourjopoulos AudioGroup, WCL Department of Electrical and Computer Engineering University of Patras, Greece http://www.wcl.ee.upatras.gr/AudioGroup/ John W. Worley

2. Slide 2 of 13 Introduction  To perceptual assess room deverberation.  Using, Complex Smoothed room transfer functions (RTF) -v- loudspeaker equalisation.  Measuring multi-dimensional sound quality attributes. F Low-frequency spectral balance. F High-frequency spectral balance. F Phase clarity. F Overall sound quality. F Room size. F Source distance.

3. Slide 3 of 13 Method: Anechoic inverse filter design  Presentation, free from loudspeaker response coloration.

4. Slide 4 of 13 Method: Complex Smoothed inverse filter design  Presentation, free from loudspeaker and RTF coloration.

5. Slide 5 of 13 Inverse filtering using smoothed filters from: “Results for Room Acoustics Equalisation Based on Smoothed Responses” Panagiotis D. Hatziantoniou and John N. Mourjopoulos,114th AES Convention, Amsterdam, March 2003 time domain frequency domain modification compensation

6. Slide 6 of 13 Test Rooms  L: 7.15m X W: 4.60m X H: 2.90m.  RT = 0.368 sec (frequency averaged).  L: 10.20m X W: 7.05m: X H:2.65m  RT = 1.1 sec (frequency averaged). Room 1 Laboratory/Listening room Room 2 Classroom

7. Slide 7 of 13  2I-multi-AFC task.  8 Subjects.  Stimuli  Music  Speech  Castanets  Snare  Exp. Conditions  On-filter -v- Off-filter presentation  Anechoic -v- Smoothed filter  2 loudspeaker pairs  2 rooms Subjective assessment

8. Slide 8 of 13 Results: Room 1

9. Slide 9 of 13 Results: Room 2

10. Slide 10 of 13  Smoothed filter rated higher than anechoic:  High-frequency spectral balance, phase clarity, overall sound quality, & room size.  Listener position = non.sig.  Sig. effect of room:  High-frequency spectral balance, phase clarity, & overall sound quality  Sig. effect of loudspeaker:  Low-frequency spectral balance. Results: MANOVA & ANOVA

11. Slide 11 of 13 Conclusions  Complex smooth filtering is not listener location specific  A preference for Complex smooth filtering shown in:  Temporal distinctiveness  Mid- to high-frequency spectral balance  Reduces reverberance  Complex smooth filtering has more effect in acoustically poor room than optimum listening room.  Listeners prefer good loudspeakers over poorer loudspeakers in terms of low-frequency spectral balance.  Agreement with objective measures.  Successful measure of multi-dimensional sound quality attributes.

12. Slide 12 of 13 Future work  Compare filtering methods to commercial dereverberation DSP techniques (ie parametric equalisation).  Measure and test Complex smooth filtering in a multi-purpose hall.  Reduce the subjective measurement variables.  Separately test the temporal and magnitude effect of Complex Smooth dereverberation.

13. AudioGroup, WCL Department of Electrical and Computer Engineering University of Patras, Greece http://www.wcl.ee.upatras.gr/AudioGroup/