Speechlab

MyDictate PDF Tisk Email

MyDictate – program for discrete-speech dictation

mydictate.jpg

The dictation tool MyDictate was designed to work with discrete-speech input. It can be used as a standalone program or as an extension to the previously developed MyVoice. MyDictate was created for text dictation, which is a slightly different type of  PC application. In this case, the active vocabulary must be very large (more than one hundred thousands items). This vocabulary changes only occasionally when the user wants to add a new word. The dictation tool must be designed to cope with this large load and it must take into account that the spoken text can be misrecognised and misinterpreted due to various reasons, like recognition errors, ambiguous items or alternate spelling, among others. Thus, MyDictate had to be equipped with features that allowed for easy and immediate correction of input text, selection from several acoustically similar (or even same) word candidates as well as any arbitrary editing of the already printed text.

A big proportion of MyDictate commands are similar to the ones of MyVoice. However, the main difference is that in MyDictate, their pronunciation form is longer than strictly needed, to improve their recognition accuracy. It is because all commands in MyDictate share the same vocabulary together with a half million of words. So for example, the command “left” in MyVoice corresponds to “cursor_left” in MyDictate. Briefly described, commands in MyDictate are prepared for:
  • activation and deactivation of the output to the OS
  • lexicon’s modification and configuration of MyDictate
  • control of cursor’s movement, editing and selection of the text and spelling of words
  • termination of MyDictate and launch of MyVoice (if these both systems are installed on the same PC)
  • changing the size of the first character belonging to the word or phrase that was dictated as the last one
  • selection of one candidate from the list of all next candidates – when this command is used then the word that was printed on the screen as the last one is replaced with the selected candidate
  • deleting a word, two words or the last character
As mentioned above, MyDictate has several important functionalities (e.g. deleting the final characters and selection of next candidates) that allow correcting (respective replacing) each misrecognized word easily. For an inflective language like Czech, this is a very important feature because in it, a lot of acoustically closed words (with the same basic form) differ only in several letters at the beginning or end. Moreover, these commands also allow that the out of vocabulary (OOV) words can be dictated easily: the user only utters their basic form at first and then he/she corrects it using commands for deleting of characters and/or spelling.


more:

  • CERVA, P., NOUZA, J.: Design and Development of Voice Controlled Aids for Motor-Handicapped Persons, In: Conference of the International Speech Communication Association (Interspeech 2007), pp. 2521 - 2524, August 2007. ISSN: 1990-9772