Abstract :
It is common for papers discussing the human factors and automatic speech recognition (ASR) to highlight potential advantages over existing computer input media. The most commonly cited of these advantages are that speech does require specialised training, that speech permits eyes free use, and that speech exploits a spare attentional resource. These points only relate to specific instances and situations. Several writers have noted that a stumbling block for well engineered ASR systems continues to be the end user. This problem may well result from designers taking the above noted `advantages´ of ASR literally, and assuming that ASR will be by definition natural, `hands/eyes free´, and capable of reducing visual workload. The author looks at these misconceptions, and considers the use of automatic speech recognition in control room environments
