求救外文翻译(关于计算机语音识别的文章)

来源:百度知道 编辑:UC知道 时间:2024/06/27 16:36:04
4.2.2. Tasks
The 28 subjects were divided into two groups and
were given a passage of hard text (331 words, reading
ease score¼40:2, reading grade level¼12) or easy
text (418 words, reading ease score¼82:4, reading
grade level¼3:8) to enter into the computer by
*speaking and typing [3]. All subjects were limited
to 5 min for text entry. Each subject performed the
following tests using their selected text:
Test1: Subjects spoke in a normal voice and were not
concerned with errors (speaking-maximum input).
Test2: Subjects spoke in a manner to cause the least
amount of error and were allowed to correct any
errors by voice command (speaking-maximum accu-
racy). For example, subjects could say ‘‘scratch that’’
to delete the prior word or phrase generated by the SR
software. Thus, no typing was needed for editing.
Test3: Subjects typed as fast as they could and we

4.2.2 。任务
28名受试者分成两组,并
获得通过辛勤文( 331字,阅读
缓解评分¼ 40:2 ,读职等¼ 12 )或容易
纯文字( 418字,阅读方便评分¼ 82:4 ,读
等级水平¼ 3时08分)进入电脑
*发言和打字[ 3 ] 。所有受试者有限公司
以5分钟为文字输入。每个主题演出
以下测试使用其选定的文本:
test1 :科目,以一个正常的声音,并没有
关注错误(讲-最大输入) 。
test2 :主题发言的方式,造成至少
金额错误,并允许以纠正任何
错误,语音命令(讲-最高精度-
情趣) 。举例来说,科目,可以说''从零开始''
删除之前,词或短语所产生的简
软件。因此,没有打字的需要,为编辑。
test3 :科目打字快,因为它们可以被
不关心错误(打字-最大输入) 。
test4 :科目打字小心和纠正任何
误差通过键盘(打字-最高
准确度) 。