With the goal to improve the Vocal Human Machine Interface (VHMI), which is expected to become a primary means of interacting with in-vehicle information equipment, we present in this paper our work to provide a system architecture for signal conditioning, before the Automatic Speech Recognition (ASR) module, which is robust to the radio sound in a vehicle environment (this disturbing sound is called "radio noise" in this paper). Classicalspeech enhancement systems are briefly analyzed and it is explained how this proposal can address their limitations. The new system architecture isfully detailed as well as the expansion possibilities. For the covering abstract see E134653.
Abstract