摘要 |
<p>A process for speech recognition in a noisy environment, in particular in a vehicle, whereby incoming acoustic signals are split up into packets of predetermined time length and fed through a frequency analyser. If these packets contain frequencies which are characteristic of human speech, the packets are compared with predetermined noise samples. The difference thus produced is compared with predetermined word patterns. If a difference matches with a word pattern, the word concerned is thus recognised. An associated control instruction is then executed in a known manner.</p> |