A Noise Suppression System for Improving Voice Recognition Rates of a Voice Activator in a Vehicle

A noise suppression system for automotive application that realizes improved signal-to-noise ratio (SNR) of voice signals both low distortion and simple system configuration has been developed.

Spectrum subtraction technology with an accurate voice signal detection technology is employed in the noise suppression algorithm.

A high-speed processing system was constructed by using a digital signal processor (DSP).

The system is adapted independently for serving as a preprocessor for a voice recognizer. A single microphone is connected to an input terminal of the system. Noise-corrupted voice signals are preprocessed thus suppressing noises in the system and then transmitting to the following voice recognizer.

Experimental results show that SNR of the voice signals could be improved 5 dB on average, and the voice recognition rate could be improved 9 % to 19 % on average with setting the system on a passenger car while driving at speed of 80 km/h to 100 km/h.