Perceptual surround sound recording and rendering

Start date April 2012

Client King's College London
Investigator

Dr David Nugent
david.nugent@elucidare.co.uk

 


Abstract

Researchers from King's College London (KCL) have invented a portfolio of emulation technologies to record and render perceptual surround sounds using computationally efficient algorithms. These techniques are amenable to a wide range of acoustic applications including mobile gaming, augmented reality, broadcasting audio, and surround sound television. Patents have been granted or filed for each design.

 

Novel surround sound microphone Coherent soundfield emulation Efficient perceptual reverberation
Physical and virtual surround sound microphones incorporating n-order differential microphones. Inter-channel time and intensity analysis improves the accuracy of spatial reproduction compared to ambisonic methods.
The impulse response of a performance venue is recorded by a circular microphone array. Post-production processing is correspondingly applied to dry audio samples to recreate the original reverberant environment.
A novel and highly efficient auralization model for sound sources within complex reverberant environments. Directional sources and frequency-dependent reflections and absorptions are supported.
US patent pending US patent granted US patent pending

 


Novel surround sound microphone

Inter-channel time and intensity dfferences are captured by a circular non-coincident array of directional microphones. Unlike the Johnston-Lam design based on hypercardioids, KCL uses customised higher order polar patterns to suppress non-adjacent crosstalk and optimise tuning to underlying psychoacoutic laws. Sound-source localisation experiments confirm superior performance of this design compared to Johnston-Lam, second-order Ambisonics, and intensity-only techniques.

The KCL design can be implemented as a physical microphone array using second and higher-order differential microphones arrays, the Eigenmike, or propriatory directional microphone designs. Alternatively it can be used in a virtual form for creating and rendering multichannel audio in arbitrary circular speaker arrangements including binaural and 5.1 configurations.

Johnston Lam KCL microphone array
The Johnston-Lam microphone array comprises five hypercardiod microphones oriented at 72-degree intervals in the horizontal plane. Two vertical shotgun microphones (not shown in polar plots) have broad response nulls at 90-degrees to minimise the amount of lateral signal pickup in those channels.

Directional patterns created by n-order differential microphones ensure sounds are detected by only two neighborouring microphones.

 

Documents available for download

US patent application, Microphone Array

Generalised design method for directivity patterns of spherical microphone array

On the design and implementation of higher order differential microphones
Panoramic recording and implementation of multichannel audio using a circular microphone array

 


Coherent soundfield emulation

This invention makes use of impulse responses of the performance venue to process a recording or other signal so as to emulate that recording having being recorded in the performance venue. In particular, by measuring or calculating the impulse responses of a performance venue such as an auditorium between an instrument location within the venue and one or more soundfield sampling locations, it then becomes possible to process a "dry" signal, being a signal which has little or no reverberation or other artifacts introduced by the location in which it is captured (such as, for example, a close microphone studio recording) with the impulse response or responses so as to then make the signal seem as if it was produced at the instrument location in the performance venue, and captured at the soundfield sampling location.

 

Documents available for download

US patent, Audio Signal Processing Method and System

 


Efficient perceptual reverberation

Digital reverberation plays an important role in computer games and virtual reality applications by increasing the level of realism. Accurate simulation of room acoustics is a computationally intensive process which is often substituted with artificial reverberators to provide a computationally simpler alternative. However, such systems generally can not accurately simulate important aspects of room acoustics such as early reflections, source/microphone directivity, and frequency-dependent absorption and reflection.

To overcome these limitations the KCL researchers have developed a real-time and scalable room auralisation methodology called the scattering delay network (SDN). Perceptual surround sounds are generated in real time - requiring low computational overhead - using readily available parameters including: the location and directivity of the sound sources, the location of the listener, the location and frequency responsivity of the primary boundary walls.

Individually the features and benefits listed below represent desirable advantages over conventional auralisation techniques. Collectively they present a radically new and improved method for implementing digital reverberation on portable devices.

SDN features

User benefit
Real-time auralisation of sounds moving within complex reverberant environments Gamer experiences an acoustic environment corresponding to the visual presentation
Exploits perceptual features to obtain good accuracy with low computational load

Suitable for computers games on mobile phones using low-power processors

SDN nodes positioned where first-order reflections originate Faithful reproduction of perceptually important first-order reflections
Frequency dependent reflections off boundary walls Improves quality of digital reverberations

 

 

Step 1: Line-of-sight audio is encoded precisely (not shown in graphic)

 

Step 2: First-order reflections are constructed precisely according to source and player geometries relative to boundaries (red balls).

 

Step 3: Second-order reflections are approximated based on first-order node positions (blue balls).

 

Step 4: Higher order reverberations are approximated based on first-order node positions (green balls).

 

 

The following auralisations are based on room dimensions and source/microphone positions according to the attached diagram. The microphone array is a stereophonic subset of KCL's patented array for base-angle 72-degree and radius 15.5cm.

Sample audio: Realistic_walls.wav

As a first sample, the walls have realistic frequency-dependent absorption. More specifically, the floor is cotton carpet, the ceiling is fissured tiles while the walls are walls, hard surfaces average.

Sample audio: Changing_reflectiveness.wav

As a second example, the walls have frequency-independant absorption, which is changed over time. For the first 5 seconds the room is perfectly anechoic. Between 5 and 10 seconds, the walls have a reflectiveness of 0.75, which then increases to 0.85 between 10 and 15 seconds. Finally, at 15 seconds the reflectiveness is further increased to 0.9.

Documents available for download

US Patent application, Electronic Device with Digital Reverberation

Scattering Delay Network: an Interactive Reverberator for Computer Games
Frequency-Domain Scattering Delay Networks for Simulating Room Acoustics in
Virtual Environments