Sound source localization techniques using microphone arrays

Localization is the detection of where, inside the environment, the sound originated from, that is its source. This involves the identification of the direction and in some cases of the distance.
Since the Seventies this subject has been a research topic and different approaches have been proposed and investigated. Several applications are possible, including multimedia systems and particularly teleconferencing, or speech recognition and speaker identification, sound capture in adverse environments (noisy or reverberant), large-room recording-conferencing, acoustic urveillance, sonar and hearing aid devices. These systems have the potential to be beneficial in several other environments. In some cases, for instance, localization techniques have been integrated with vision, using in this way data provided from two different senses.

Systems developed in order to solve the localization problem use an array of microphones, because it has potential capabilities that a single microphone does not. It may be electronically aimed to provide a high-quality signal from a desired source location while simultaneously attenuating interfering talkers and ambient noise This case is also called spatially selective sound capture. An array system does not necessitate local placement of transducers or encumber the talker with a hand-held or head-mounted microphone, and does not require physical movement to alter its direction of reception. An essential requirement of these sensor array systems is the ability to locate and track a speech source. Location data may be used as a guide for discriminating individual speakers in a multi-source scenario. With this information available, it would then be possible to automatically focus upon and follow a given source on an extended basis.
Recently the application of video-conferencing systems became of particular interest. In this regard, the pointing of an automatically voice-activated camera may allow to eliminate the need for human operators (usually participants have to control it manually). This kind of solution is the most natural and does not require any effort from the users, because it could be both burdensome and distracting.

This work is divided in two parts. The first one is about the description of the subject from a theoretical point of view and the second one regards the practical tests. In the present chapter the applications treated in literature have been already described, moreover microphone array systems used in this kind of context have been discussed. Among all these strategies, two of them have been considered of particular importance and they are presented more accurately, showing also the associated mathematical background. They are based on Time Difference Of Arrival (TDOA) and spatially selective sound capture. These techniques are considered very important because of the wide use and the fundamental role from a historical point of view. The first of these approaches have been just applied in the experiments, which are described in detail in the second part.

Here is a short summary about the content of each chapter:

- Chapter 1 introduces briefly the work carried out.
- Chapter 2 exposes the main methods used to solve the localization problem, beginning from the way to face it in humans.
- Chapter 3 shows the instruments which have been used and their setup during the experiments. Particularly important among them is the dynamic signal analyzer, because of the amount of time dedicated to familiarise with it and to test how it works.
- Chapter 4 describes the experimental tests executed inside this project and the results which have been achieved.
- Chapter 5 conclusions are drawn about the work, exposing results and possible developments for the future.