Thursday, October 13, 2005

Microphone Arrays—The One Thing from Star Trek You Can Use

Speech recognition has come a long ways in the last 20 years. From crude tools in the early 1980s to more sophisticated systems today, we are coming closer to that Star Trek-like experience where you don’t have to carry microphones or type into a computer, but rather the computer keeps track of you and operates based on your voice command. In this blog the author describes the history of speech recognition and talks about the coming generation of tools called Microphone Arrays.

Microphone arrays use multiple microphones at different locations for pinpointing the location of sound. It can be used for more effective speech acquisition because it can provide speaker location as well as hands-free speech acquisition. For an in-depth review of the technical background for Microphone Arrays, please click here.

Applications for microphone arrays include cellphone based telephony, video teleconferencing, and audio interfacing with PC systems. Here’s an example from TNO in the Netherlands applying the technique to a surgical operating room. The entire wave field from the room is scanned. The system suppresses unwanted signals and amplifies the desired signal. Video conferencing is another key application. Polycom makes a microphone array that wouldn’t look out of place in the office.

Research is well under way at numerous universities. Here’s an example from Rutgers, and one from Brown University. The NRC in Canada has a Beam forming project in which an array of microphones is aimed at a subject in order to track his location unencumbered with a microphone and capture audio from the subject. MIT has the largest microphone array system consisting of a 1000+ node array of microphones for a research project called Loud (Large acOUstic Data array project).

BMW uses microphone arrays implanted in the helmet for motorcycle drivers. They use two microphones placed near the forehead of the driver which allows them to capture the audio waves from the driver’s face. You can see their implementation here.

Advances in silicon chips make microphone arrays possible. Check out the DiSiMic from Sonion which is an ASIC with a Sigma-Delta converter and a low-noise preamp. It generates a digital output and is targeted to the mobile phone market.

If you are working with Microphone Arrays, I would like to hear from you. You can reach me at hall.martin@ni.com.

Best regards,
Hall T. Martin