Paper: "Identifying Principal Attributes for Evaluating Audio Quality of Reproduction Systems With Spatially Dynamic Program Material"

[MaxwellsEq]

Hi all, this is an AES paper (freely available online!) from Pia Poryse a PhD student Aalborg University and Bang & Olufsen. It uses an interesting methodology to identify the basic audio quality of sound replay systems (level matched) and to use this to identify whether moving sound images (spatially dynamic) has an impact on people's ratings. The conclusion seems to be that reproduction quality dominates the results, but there is a small measurable impact that may change preferences based on spatially dynamic content.

Abstract: Despite the significant advances in spatial audio technologies, there is a lack of knowledge regarding the impact of spatially dynamic program material on known descriptive attributes of basic audio quality (BAQ). This paper presents an exploratory study that investigates the perception of sound quality in reproduced sound using spatially dynamic program material in comparison with stationary program material. Three reproduction systems (stereo, surround, and surround with height) were compared across two different spatial variations of the same complex audio scene. Using a paired comparison approach combined with the check-all-that-apply (CATA) method, differences in BAQ ratings were collected alongside attributes contributing to the decision of BAQ rating. The study reveals that audio quality and its underlying attributes are primarily impacted by differences in reproduction systems, with spatially dynamic program material showing no direct influence. However, there is slight evidence indicating that the use of spatially dynamic program material may influence the selection of attributes for comparisons of different reproduction systems, leading to less clarity in results. Furthermore, this study can serve as a practical guide as it lists attributes for various loudspeaker comparisons and contextualizes their impact on quality ratings.

Identifying Principal Attributes for Evaluating Audio Quality of Reproduction Systems With Spatially Dynamic Program Material

 

[Sokel]

Only ten participants though.
(Genelec 8320A and 8331A were used,they must read ASR )

 

[Blumlein 88]

Only ten participants though.
(Genelec 8320A and 8331A were used,they must read ASR )

Actually they discarded 3 for much of it. So seven participants. My first blush is this being an example of if you tease the data enough something will show up. OTOH, the attributes most noted as different were just about exactly what you would expect comparing stereo to surround. And surround with height and without. Plus the addition of height did not seem to matter all that much when comparing surround with and without height.

 

[Blumlein 88]

The conclusion seems to be that reproduction quality dominates the results, but there is a small measurable impact that may change preferences based on spatially dynamic content.

I did not interpret the results quite this way. It seems spatially dynamic material scattered the scoring of attributes. That would include changing attributes chosen, but not in a consistent way. In a way that muddles which attributes get scored more often. Attributes that had some consensus among listeners had much less when the signal was spatially dynamic. I would not call that changing preferences, but scrambling them. It would indicate to me that spatially dynamic signals reduce the ability of listeners to consistently judge sound attributes.

From the conclusions in the paper:

Generally, when dynamic audio is introduced, there is a reduced ability of the panel to clearly determine the attributes that are affecting their decisions, evident through a more uniform distribution and fewer attributes being selected more frequently than expected.

 

[Blumlein 88]

Here is a paper testing CATA (check all that apply) attributes for speaker comparisons. I wasn't familiar with that method in audio prior to seeing it today. So this fills in more about it.

Sensory Profiling of High-End Loudspeakers Using Rapid Methods—Part 3: Check-All-That-Apply with Naïve Assessors​

https://independent.academia.edu/BechSøren?swp=tc-au-84419267

Sensory Profiling of High-End Loudspeakers Using Rapid Methods—Part 3: Check-All-That-Apply with Naïve Assessors

Sensory Profiling of High-End Loudspeakers Using Rapid Methods—Part 3: Check-All-That-Apply with Naïve Assessors

www.academia.edu

 

[MaxwellsEq]

I did not interpret the results quite this way. It seems spatially dynamic material scattered the scoring of attributes. That would include changing attributes chosen, but not in a consistent way. In a way that muddles which attributes get scored more often. Attributes that had some consensus among listeners had much less when the signal was spatially dynamic. I would not call that changing preferences, but scrambling them. It would indicate to me that spatially dynamic signals reduce the ability of listeners to consistently judge sound attributes.

From the conclusions in the paper:

Generally, when dynamic audio is introduced, there is a reduced ability of the panel to clearly determine the attributes that are affecting their decisions, evident through a more uniform distribution and fewer attributes being selected more frequently than expected.

Sorry, my summary was poor. I was trying to be succinct, but in the end I oversimplified it .