Digital Signal Processing Reference
In-Depth Information
Chapter 1
Introduction
1.1 General Principles of Microphone Arrays
Ever since the invention of telephone systems, sound signal acquisition has
become an essential part of acoustic and speech processing. Traditionally,
sound signals are recorded and processed with the use of a single microphone;
but this way of acquisition suffers significant limitations: not only the sound
spatial realism is lost but also the flexibility in processing the recorded signal.
Indeed, extracting the signal of interest from the noise, reverberation, and
competing sources, is confined to a small family of techniques. To preserve
the sound fidelity, spatial realism, and increase the processing flexibility in
dealing with multiple sources and noise, the concept of microphone arrays
has been introduced [1]-[5].
Simply put, a microphone array consists of a number of acoustic sensors,
which are organized together to sample the sound field with spatial diversity.
As a result, the array outputs contain the signal of interest, noise, inter-
ferences, and also the propagation information that is represented by the
acoustic impulse responses from the radiating sources to the microphones.
Through processing the array outputs, various functionalities can be im-
plemented including but not limited to: localizing and tracking the sound
sources, extracting the signal of interest, suppressing ambient noise, separat-
ing different sound sources, and recording spatial sound. The degree of how
well these functionalities are accomplished, however, is conditioned on many
factors such as the quality of each individual microphone, the number of
sensors, the array geometry, the operating environment, and the processing
algorithm (called beamforming). The difference can be enormous, depending
on how the array is designed. Therefore, a considerable amount of attention
has been paid to the design issues of microphone arrays and many different
types of them have been developed over the last few decades. Broadly, those
arrays can be categorized into two basic classes based on different principles:
additive and differential arrays.
