Mrosek B, Grünupp A, Keppel E, Kunzmann S, Moese G, Mohr K, Stargardt A, Günther R W
Klinik für Radiologische Diagnostik, RWTH Aachen.
Rofo. 1993 Nov;159(5):481-3. doi: 10.1055/s-2008-1032802.
A voice recognition system, called Tangora, has been designed for the German language; its prototype has been tested to assess its clinical usefulness for radiological reporting. Its technical vocabulary consists of 20,000 words. The system recognises words spoken provided short pauses are inserted. The Tangora system was compared with conventional methods of dictation, using 68 radiological reports. The work necessary for each system was analysed. The average time taken by the Tangora system per report was 671 seconds compared with 182 seconds for conventional dictation but improvements in the speed of the new system for radiological reporting is to be expected.
一个名为Tangora的语音识别系统已针对德语进行设计;其原型已通过测试,以评估其在放射学报告中的临床实用性。其技术词汇量为20000个单词。该系统能够识别插入短暂停顿后说出的单词。使用68份放射学报告,将Tangora系统与传统听写方法进行了比较。分析了每个系统所需的工作量。Tangora系统处理每份报告的平均时间为671秒,而传统听写为182秒,但预计新系统在放射学报告速度方面会有所改进。