Spatial-Sound Student Projects

What is this all about?

The CIPIC Interface Laboratory has identified a number of interesting projects involving spatial or 3-D sound, and is looking for undergraduates and MS students who would like to work on them.

Why is spatial sound important?

The electronic reproduction of sound for entertainment has evolved from the familiar stereo systems to multichannel surround-sound systems. Major advances in quality and realism for entertainment systems and for applications such as human-computer interaction or wearable personal communications are still to come. These applications will need sounds that not only seem to come from left or right, but also from above or below, and from near or far. Future systems will rely on an improved knowledge of the human spatial hearing system and on the digital signal processing that exploits it.

Where is this research being done?

This work is happening at the CIPIC Interface Laboratory, which is engaged in program of basic and applied research and development in spatial sound under the sponsorship of the National Science Foundation and of California industry. This research program offers opportunities to good students to get involved in important and exciting projects.

What do I need to know to do one of these projects?

The general background needed is personal interest in audio and good software and analytical skills. Also of importance for some projects are a solid knowledge of signals and systems and digital signal processing, with skill at using MATLAB being desirable. There is a comprehensive tutorial on spatial sound on our web site, if you really want to dig into the details. Our website also includes some demos.

To understand and get started in the projects, it is enough to know the following basic facts. A sound that originates at a location near a person is modified on its way to the ears by the room, walls, ceiling and furnishings, and also by the listener's own body. Thus, the sound pressure at each person's ear drums depends on the body of the listener and on the acoustic environment. In free space, or in an anechoic room, the Fourier transforms of the sound pressures at the two ear drums, Y₁ and Y₂, are related to the Fourier transforms of the sound pressure of the sound source X by the transfer functions H₁(f) and H₂(f) of two linear systems. These are called the head-related transfer functions or HRTFs, for short. The HRTFs are functions of frequency, and change depending on the location of the sound source relative to the head. In other words, there is not a single transfer function, but a large set of transfer functions, one for each direction that sound comes from. Finally, it is important to know that there is no universal HRTF. Just as each person has his or her own facial features or size, each person has his or her own HRTF. We have developed sophisticated methods for measuring HRTFs experimentally, and we have several ways to approximate them by signal-processing models such a combination of time delays and pole/zero filters. A very good rendition of spatial sound can be achieved over headphones with signals that replicate what the individual listener would perceive for live sounds in a real room.

Project descriptions

The projects described below provide some examples of possible student projects. Most of these projects can be done by one student or by a team of two students. They range from basic research to applications and demonstrations. The brief descriptions are merely intended to give an idea of the kinds of topics of interest; you should see us if you would like to know more about exactly what is involved.

A. Basic research in spatial sound

Although much is known about HRTFs, there are still some important unsolved basic problems that need serious attention. For example, it is experimentally difficult to measure HRTFs at low frequencies (say, below 500 Hz). We have reason to believe that people are sensitive to low-frequency cues, but this topic is still far from being well understood and needs to be thoughtfully investigated. For another example, we know that HRTFs change with body posture -- they change when you turn your head, or when you shift position from being seated to standing. However, we don't know when these changes have a strong effect on perception, or what to do about adapting to posture changes in a spatial sound system. There are other unsolved basic problems that arise when we attempt to provide useful computer output through spatial sound instead of through text and graphics. Tackling these basic research questions requires a desire to investigate and a willingness to persist when no clear solution exists. The problems are challenging, but their solution can lead to major advances in spatial sound systems.

B. Laboratory based applications and demonstrations

The CIPIC Interface Laboratory has established advanced capabilities for rendering high quality spatial sound either on headphones, in stereo, or on a Dolby 5.1 sound system, and for tracking head motions in real time. These capabilities provide the technical environment and the opportunity for interesting applications, such as:

1. Using spatialized synthesized speech to provide computer output
2. Using head gestures for interacting with spatial audio output
3. Using head tracking to greatly improve spatial illusions created by loudspeakers
4. Developing a theater-sound-quality headphone-based home theater system
5. Implementing an immersive concert -- a musical performance where instruments can be located flexibly about the listener
6. Developing spatial audio games

C. Wearable computers and mobile demos

We are also interested in the research and development of advanced spatial sound capabilities for mobile computers and portable devices. This requires the processing of sound signals with the personal HRTF of each user and the measurement and use of the important cues provided by the motion of the head. All these capabilities are to be implemented on a small portable system that may combine locally processed information and the downloading of data over a wireless network. An interesting hardware/software project that has already been started is the design and implementation of an inexpensive untethered head motion tracker. Although the development of fully functional complete systems may not be feasible, there are numerous opportunities to develop interesting and important components of such a system.

I am definitely interested. What do I do next?

To pursue this further, send an e-mail message to Prof. V. R. Algazi identifying the project(s) of interest and some information on your background and preparation.

Last revised:11/8/2001