Investigating how height delivered using Ambisonics impacts the listening experience
Project from - present
What we're doing
We are investigating one of a number of techniques that could deliver the next generation of surround sound. Ambisonics, which was invented in the 70s, is capable of delivering with-height (or '3D') audio to many different speaker layouts using only one set of signals. This project is looking at how such a system can best reproduce with-height audio and what impact the height has on the listening experience.
Why it matters
Surround sound has been standardised and a vast amount of research has been carried out into the perception and the benefits it brings over stereo formats. Since viewing has been enhanced and moved into high resolution it seems that audio has been neglected. It is therefore necessary to investigate ways of enhancing the listening experience, and find new ways in which we can make this experience more exciting and better for the listener, however at the same time still keep systems manageable for broadcast and storage. It is necessary therefore to quantify the benefits for the user by carrying out a thorough investigation of the user experience, combining subjective evaluation and objective measures.
- Indentify the optimum reproduction resolution with respect to human hearing
- Investigate viable with-height speaker layouts
- Identify the improvement height brings over current formats
- Explore the user experience in terms of subjective attributes
How it works
Over the past decade there has been a large amount of research into surround systems and it has been shown that there are a number of attributes which make up a subjectively pleasing surround reproduction, one of these being the ability to localise sources. However, in terms of with height systems, there is not a lot of data available in this respect. Since Ambisonics provides different resolutions of reproduction at the expense of increasing speaker numbers, and since human hearing is not as accurate in the vertical as in the horizontal, localisation tests have been carried out. These will determine if there are significant differences between various Ambisonic orders for elevated positions around a listener, and inform as to whether higher order material is needed for elevated sources or not.
Another important attribute in surround reproduction is spatial impression which can be broken down into two constituent components - apparent source width and envelopment. Since there is no standard layout for surround systems with height, the next stage is to investigate a number of different layouts subjectively and objectively to find the optimum configuration to produce an enveloping sound-field. A number of different speaker configurations have been modelled and tested using an objective measure (inter-aural cross correlation) to find the optimum layout. The next stage will be to execute a subjective evaluation of the most promising layouts and collect subjective data, which will then lead to a more in-depth evaluation.
To date, two localisation tests have been carried out. The first test evaluated sources placed in the median plane (which locates sources symmetrically with respect to the listeners ears) using pink noise as the test source. The outcome of this showed that there was no significant difference between orders in terms of localisation error.
The next localisation test placed sources off the median plane, this time using two different sound sources, pink noise and female speech. This test found that there were significant differences between 1st and 3rd order Ambisonics using the female speech item in terms of localisation error, demonstrating that lower bandwidth material may require higher resolution. The results were published at the Reproduced Sound conference 2012.
Moving on from the resolution of a system another important part of a spatial audio system is how well it can portray spatial impression. To this end a number of different virtual systems (horizontal and vertical) have been simulated using HRTF’s, these have been measured using IACC and it has been found that there is a notable difference between a number of systems, further a trend has been identified which matched that of current studies by NHK, in terms of spatial impression.
This project is part of the Immersive and Interactive Content section
This project is part of the Audio Research work stream