Periphony for Broadcast
Developing the next generation of surround sound
Investigating whether Ambisonics can present improved off-centre listening quality for next-generation surround sound systems
What we're doing
In the field of spatial audio reproduction broadcasting, there are a vast amount of elements that are combined in order to record, store, transmit and reproduce a sound field that is pleasing to the user, accurate and in-line with artistic intent. There are a number of challenges that each element of the reproduction chain introduces and in order to try and alleviate these challenges, they must first be understood. Through a detailed review of the literature, it has been found that one key challenge of spatial audio reproduction systems is the ability to reproduce a sound field accurately over a large listening area. This topic has research based on many different reproduction systems, however, Ambisonics has highlighted significant proposed solutions to this problem including an increase in sweet spot size relative to Ambisonic order, and also the use of sweet spot specific decoders. This project aims to assess the problems with both accurate reproduction over large listening areas and also the tools and methods used to assess these metrics. These measures will be acquired through modelling, objective assessment and subjective testing with an emphasis on perceptual psychoacoustics.
With close collaboration with BBC R&D, initial investigations into psychoacoustic modelling of non-central listening positions was undertaken. In recent developments and using our initial modelling as a guide, we have created a dynamic binaural auralisation system using optically tracked markers to auralise any spatial audio system using stringent control over listening conditions (in this case, we can control the listening position by the choice of HRIR database and virtual loudspeaker driving signals). This means that purely blind testing of non-central listening positions can be achieve, with accurate dynamic interaction between the listener and the soundfield. This system will be used to complement objective measures and modelling. Working closely with Chris Pike, we have developed and tested the system and also planned for further measurements in order to achieve non-anechoic HRIRs for improve externalisation.
Localisation Performance of Higher-Order Ambisonics for Off-Centre Listening
More project info
Why it matters
We want to assess whether Ambisonic based spatial audio system can present an improvement over currently used spatial audio system (5.0 based using amplitude panning) and the effect Ambisonic variables has on the size and characteristics of the listening area. Previous research into Ambisonic based systems have proposed that there is a relationship between Ambisonic order and the size of the "sweet spot" (region of accurately perceived listening) but very few researchers have successfully proved this relationship and many of the research publications looking into the assessment of any spatial audio reproduction system often concentrate on the optimal listening position (equidistant from all loudspeakers). The global goal being to provide a reference which will in-turn improve user perceived quality.
- Use psychoacoustic metrics and relevant modelling to compare localisation at central listening positions.
- Identify the reliable and proven methods of approximating listening area characteristics.
- Implement a reliable listening area size prediction method (from the assessment in 1 & 2) to approximate a clear definition of sweet spot size as Ambisonic order changes. Subjective testing would aid in research solidarity.
- Perform a sweet spot comparison for currently used reproduction systems especially including the ITU 5.0 layout.
- Identify what the effect of Ambisonic reproduction system variables (loudspeaker layout, decoder design, encoding techniques, number of dimensions) have on listening area characteristics.
- (time permitting) From the listening area assessment method, create and test a decoder matrix iteratively using the confinements of acceptable accurate listening area size.
- (time permitting) Consider irregularity compensation using psychoacoustic and listening area assessment methods defined above.
How it works
Initial objective models have been undertaken by comparing the low-level localisation cues (ITD and ILD) synthesised by a number of spatial audio reproduction systems at various listening positions. Following this we are now in the process of developing and validating a testing methodology for off-central listening positions using a dynamic binaural auralisation system. Once validated, this system can be used to undertake blind subjective assessments of various spatial audio systems with control over the listening scenario - by using measured or synthesised HRIRs at non-central listening positions, we can create a blind test rarely achieved for multiple listening conditions.. The results from this experiment will then be compared against further objective modelling to assess the relationship between Ambisonic order and listening area characteristics, and also how Ambisonic systems compare to the current industry standard spatial audio reproduction systems.
- Implementation of Ambisonic decoder with commonly used objective analysis (energy and velocity vectors)
- Analytical modelling and low-level psychoacoustic analysis at central listening position and non-central listening positions (papers presented at International Symposium on Modern Acoustics 2012 and Institute of Acoustics: Reproduced Sound 2012).
- Analysis of the subjective listening-area quality assessment methods
- Development and validation of a real-time dynamic binaural auralisation system (currently 2D only)