A speakerphone is a telephone that operates in a hands-free mode. Ambient noise tends to cause the speakerphone to electronically switch to the transmit mode, thus "killing" the received signal from the other user. This is an inherent problem in all speakerphone designs. It is desirable to eliminate this "false switching" problem as well as reduce the ambient noise that is transmitted through the system.

This research investigates the use of a single multiple-input neuron to cancel ambient noise from the speakerphone's transmitted signal. The primary microphone samples the original speakerphone signal, which includes voice and ambient noise. A secondary microphone is added to the system to sample ambient noise. The approach is implemented in an 8051 microcontroller and will be tested in a commercial speakerphone housing with various external ambient noise sources such as computer fan noise and dot-matrix printers.
 

• (No web site available)
• Return to 1998 Project List
• Return to main Senior Project page