Image Processing Reference
In-Depth Information
11.3.3 Software Design
The method of immersive sound field reconstruction can be implemented more
effectively with object-based audio signals. Object-based spatial audio is a concept
of delivering audio which operates by rendering each audio object that appears in
the auditory scene [
12
]. In WFS, the auditory scene is created by rendering the
sound field within a given listening area corresponding to the captured scene. The
rendering module is responsible for distributing the sound objects within the
generated sound field according to the scene description parameters.
Each input signal to a driver is defined by WFS algorithm which computes
signals for the predefined virtual audio objects. The main process of DSP is
illustrated in Fig.
11.6
. The data loaded from the input consists of the audio signal,
source information, sampling frequency, and so on. The source information
includes the spatial position vector of each virtual sound source. In addition, the
distance from the loudspeaker array to the listening position, the number of drivers
and a gap between them are given in advance.
The first step of the algorithm is the calculation of the distance between each
driver and the virtual source. The driving function calculates the time for each
sample that the sound front takes from the virtual source arriving to each driver,
based on the input data. Then the convolution of the output driving signal is
performed with the impulse response of each driver. The signal after the convolu-
tion is described in the frequency domain through the FFT process so that the
directivity can be included. The directivity of each driver according to the fre-
quency and angle of emission is measured before the WFS algorithm starts.
Generally, it can be measured in advance by the calibration process. The frequency
contents are transformed back into the time domain via the inverse FFT process.
The output vectors can be saved as audio files.
Fig. 11.6 Block diagram
of main software
