87 A Microcontroller-Based Neural Network for DTMF Decoding
Download citation file:
In an earlier paper we demonstrated that reliable dual-tone multifrequency (DTMF) decoding can be accomplished using feedforward neural networks. Many possible solutions for DTMF decoding have been successfully employed, including filter banks and FFT algorithms. These designs often increase the parts count and complexity of the systems. Software designs can require manipulation of complex numbers, resulting in a heavy burden for integer-based microprocessors. This paper extends our work by addressing additional system noise for the PC implementation, and by introducing a microcontroller-based decoder using an inexpensive PIC 18F452.