The Jess & Mildred Fisher College of Science & Mathematics

School of Emerging Technologies

Voice Writing Center for Radio Captioning

Ellyn Sheffield (Psychology), in partnership with Michael Starling (National Public Radio)

Since radio's inception, individuals with hearing loss have not had access to live radio programming. In order to make captioned radio a reality, Towson University and NPR (National Public Radio) joined forces, creating the International Center for Accessible Radio Technology (ICART). As radio has migrated to digital distribution, over the past four years ICART has proven the technological feasibility of such a system. ICART conducted a successful test of the radio captioning system on Election Night 2008, enabling deaf and hard-of-hearing audiences in five cities to read live coverage of the presidential election presented by NPR.

Before a full launch is possible, ICART must ensure all of the challenges unique to captioning for radio have been addressed. Closed captioning for radio does not merely mimic captioning for television for several reasons. First, the quantity of radio programming is enormous. Nationally distributed public radio comprises nearly 120 hours per week of most listened to programs as well as around-the-clock local programming from nearly 1,000 public radio stations. Second, the quality of radio captioning is critical. Closed captioning has been around for decades but has been used primarily for television and other visual media. These media require captioning to support a story-line rich with visual cues, but radio does not follow this model. Finally, real-time captioning in the U.S. for television has traditionally been done by relatively expensive, highly skilled stenographers — a system that is not feasible for the volume of live radio content being produced. Both national and local broadcasters require an inexpensive, accurate way of providing captioning in their respective markets.

A new system called “voice writing,” now in use for television captioning at the BBC (Great Britain) and NHK (Japan), addresses some of these challenges. In voice writing, trained announcers re-speak what they hear over audio feeds into voice recognition software which converts the speech into text. Although these systems are in their infancy in the U.S., they have recently had increasing success. Voice writing may prove to be a cost-effective solution for radio captioning. Successful implementation of closed captioning for radio using voice-writing technology is the last hurdle before a nationwide roll-out can occur. As part of the ICART agreement with NPR, Towson University has committed to exploring the establishment of a voice-writing center.

Impact on Students

Three graduate students are instrumental in this project. Elizabeth Moore (Experimental Psychology) is the lead voice writer and editor for radio captions; her research is in the area of non-speech information (NSI) and ways to include this information in the caption stream. Jenna McKenzie (Speech Pathology) and Jessica Barry (Experimental Psychology) are speech writers/editors, who perform editing and transcription for captioned videos as well as for radio captions

External Parnters
In The News
Related Work
  • Ellyn G. Sheffield, Michael Starling, and Daniel Schwab. (2011). Bringing Text Display Digital Radio to Consumers With Hearing Loss. J. Deaf Stud. Deaf Educ.16 (4): 537-552. doi: 10.1093/deafed/enr011
  • Ellyn G. Sheffield, Dan Schwab, and Mary Hinch. Radio Text Displays for Radio Captioning: Results from Three Studies. A report prepared for National Public Radio. (.pdf).






University Police

Closings & News
410-704-NEWS (6397)

Text Alerts
Sign up now

Photo Gallery