Use of Dragon NaturallySpeaking for Real-Time Transcription for the Deaf or Hard of Hearing

Overview

Generally, there are two methodologies for using Dragon NaturallySpeaking for real-time close captioning. In the first, a speaker has a microphone which is broadcasting to a computer which has Dragon NaturallySpeaking running which is doing the real-time transcription. In the second model there is a redictator who is listening to the speaker and is dictating their spoken words which broadcasts to the computer.

In the first model, there are several implementation issues that will assist in optimizing the effectiveness of this approach to real-time closed-captioned. Ultimately, the throughput accuracy will vary greatly depending on the speaker, the environment, and the subject being discussed. It is our opinion that only the deaf/hard of hearing community can determine whether the general throughput accuracy generated in any given situation is adequate for their needs.

In the second model, the primary advantage is that the dictator can be skilled at simultaneous dictation which can result in much higher throughput accuracy; but the obvious disadvantage is that this requires another person, as well as associated costs.

Model One: Real-Time Closed-Captioning

  1. There are various wireless input microphone modalities, including the lapel x-Tag mic from revolabs and the Jabra 9330, which, while not as small as some microphones, position the microphone element appropriately for optimized accuracy.
  2. Prior to commencing real-time captioning if it is possible the user should create a Dragon NaturallySpeaking profile. If this is not possible one can maintain a generic adult male and adult female profile on a laptop and use these for transcription (the overall throughput accuracy using this might be reduced to a point that it is not valuable for its purpose).
  3. There are several programming tools available in the NaturallySpeaking products starting with the Preferred product line which can "preprogram" the context in which the speaker is going to be speaking. This means that someone will be doing an ad hoc presentation on a technical subject can preprogram and all of the technical term so that the overall accuracy will be much higher than that this tool was not used.

These tools include:

  1. Add words from documents tool in the Accuracy Center
  2. Vocabulary Builder tool
  3. VocTool-located from the Dragon NaturallySpeaking menu in Professional
  4. Obviously, an ad hoc real-time close captioning the speaker will not be using punctuation.  The resulting output text will be generated in the form of "stream of consciousness". This is applications not only from a throughput accuracy standpoint, but also from a reader standpoint.
  5. Wherever possible, the speaker should be coached prior to doing a presentation with this technology on clarity of diction and understanding what makes speech recognition technology operate at its best. If feasible, it would be

Model Two: Remote Re-dictater

In this model a wireless broadcast microphone is used to broadcast to a listener either in the same location or in a remote location. This listener is skilled in simultaneous re-dictation which can be captured and transcribed on their PC, while simultaneously connected to a student's computer (laptop) who can view the transcription in real-time as the simultaneous re-dictater generates the text. This is similar to a 'simultaneous translator' as one might find at the United Nations. There are many features that recommend this model; particularly as it offsets some of the limitations of state-of-the-art speech recognition technology when used for real-time closed-captioned. In addition, it allows for the ReDictater to be capturing and transcribing classroom discussion and question and answer sessions. By placing the ReDictater in the remote location there's no "distraction" for other members of the classroom or meeting.

There are quite a number of wireless technologies that can be used to broadcast the instructors dictation or capture a classroom discussion. In a pilot program testing this concept a simple screen sharing software application was used so that a deaf hard of hearing student in the classroom was able to view the ReDictater's display in real-time. With the proliferation of remote access technologies this is now made even easier.

We would be happy to assist anyone looking to pilot speech recognition technology under either of the above models. Please fell free to contact Edward S. Rosenthal, President and CEO.


Copyright 1999- 2010 Next Generation Technologies Incorporated

Jump to: site navigation | page content

Ways Dragon NaturallySpeaking can help the Deaf or Hard of Hearing:

Highlighted Products and Services

Site Navigation