Singapore Team’s "Surround Sound" Audio Vies for Texas Instruments $100,000 Prize
A strong urge to convert the audio that is retrievable from web sites into fully immersive sound led two Singapore students to develop a winning entry that enables "surround sound" quality from just two speakers in the finals for Texas Instruments’ DSP Solution Challenge. The team of two students from Nanyang Technological University (NTU) in northwest Singapore already has won $10,000 from Texas Instruments in their quest for the $100,000 grand prize.
Siew-Cheng Eng, 27, an electronics major, and Chee-Weng Toh, 26, a communications major, both of whom are in their final year of undergraduate studies, have developed the Dual-Channel AC-3 system. This project converts the 5.1 channel Dolby Digital™, also known as audio coding (AC-3), sound into two channels, enabling a user to enjoy audio with the "surround sound" effect from a pair of speakers or a set of earphones.
Besides traditional audio systems including automotive sound systems, their project could find uses in multimedia. such as that found in personal computers, where adding more speakers would be costly and require too much space. No longer would music buffs wishing to enjoy the latest hits be limited to hearing only two stereo channels from the speakers accompanying a PC. Rather, the two speakers, equipped as tested by the contestants, would fool the ear and the listener’s brain into thinking the sound came from around the room, a sound normally produced by six speakers.
How it Works
"We wanted to find out whether we could trick the ear into believing that the sound source is coming from some locations around the head using only a pair of earphones or speakers," said Eng. The students embarked with three goals: first, to maintain original performance to the 5.1 channel system; second, to make it programmable for versatility, and third to make it affordable by the general public.
"The project works by taking in the five full-range channels (front left, front right, center, and "surround sound" rear left and rear right) as well as a sixth bass-only effect channel (thus, 5.1 channels) in the AC-3 data streams," Toh said. Then, the team used a TI TMS320C542 DSP to process information based on complex mathematical models of how people hear. The mathematical function is unique for each position and responds differently for each side of the head.
"With the software, the team can simulate the sound at any position around the listener’s head, then mix it in real-time or instantaneously," Eng said. DSPs also are used for cross-talk cancellation for the two speaker solution.
"Using the 16-bit fixed-point DSP, we are able to reduce the cost of the DCAC-3 system without sacrificing the performance of the Dolby AC-3 too much," Toh said. The team measured performance by testing 10 listeners wearing normal headsets who identified the source locations of audio heard.
The Singapore team found that using software to perform 3D audio positioning has compelling advantages in terms of flexibility, scalability, easy upgrading and cost when compared with hard-coded algorithms. The hardware platform designed, in addition to AC-3, could be used for realistic placement of speakers in multi-conversation teleconferencing; synthesizing a virtual concert or expanding a mono-sound source into pseudo-stereo. It could be used to position audio sources from many devices such as phone, fax, background music, fire alarm and e-mail within a headphone device, allowing users to prioritize action to the incoming signal sources.
"It also could be used for video arcade games when the illusion of sounds occurring around the head could add excitement for the player," notes Toh, whose hobby is PC gaming. Eng enjoys volunteering in the community in his spare time.
The students were advised by Dr. Woon-Seng Gan, senior lecturer in the School of Electrical and Electronic Engineering at NTU. Dr. Gan earned a bachelor of engineering in 1989 with first class honors and his doctorate in electrical and electronic engineering in 1993 from the University of Strathclyde, United Kingdom. His DSP experience includes the development and implementation of adaptive channel equalizer, noise and echo cancellation and audio applications both at the Center for Parallel Signal Processing, UK, and at NTU.
Last year, another team from NTU won the grand prize in TI’s DSP Solutions Challenge. Dilip Krishnan and Showbhik Kalra, and advising professor Man Nang Chong produced a restoration process for classic motion pictures and family home movies that used parallel processing techniques on a network of DSPs.
TI's DSP Solutions Challenge is a worldwide competition to develop the most innovative and functional design using TI DSPs. Created to address the industry-wide shortage of design engineers with DSP experience, TI has developed and invested in this international recognition system to encourage research and interest in the technology.
The DSP market is growing at an average annual rate of 30%, according to the market research firm Forward Concepts. Today's DSP solutions market is roughly $5 billion and has grown more than 40% per year since 1988. This market growth projection exceeds that of the semiconductor industry in general.
Related Documentation: