|
...result in marketplace advantages for the enterprises, agencies, or universities that pay attention to such requirements. In order to improve accessibility for persons who are deaf and hard of hearing, mechanisms to transform audio into other forms are needed. Automatic speech recognition (ASR) provides one such audio conversion mechanism, but there are still many obstacles to full access using this technology. In this paper we present a number of speech-to-text related solutions and provide recommendations for future enhancements.
This paper is organized as follows. First, general issues regarding access for individuals with disabilities, particularly those who are deaf and hard of hearing, are presented, along with current approaches for improving access. The Liberated Learning Project is discussed in detail as a model scenario intended to use speech technologies to better engage students with a range of disabilities. The technology supporting the Liberated Learning effort is ViaScribe *, and details of ViaScribe features and functionality are included. We then discuss CaptioneMeNow, a system intended to provide on demand, semiautomated transcription of audio on the Web. Finally, a number of future developments for ViaScribe as well as overall accessibility enhancements are presented.
Accessibility in academic and business settings has increased in many dimensions over the last decade. Accessibility options based on IT (information technology) have become more readily available and more affordable. Legislation in many countries has created compliance mandates for businesses, institutions, and agencies, resulting in a higher prevalence of accessible infrastructures. Violations and exceptions abound, but there is increased awareness of the need to make buildings and IT systems accessible for users with a range of disabilities.
Transcription and sign interpretation of audio are critical access points for users who are deaf and hard of hearing. Approximately 22,000,000 Americans are listed as deaf or hard of hearing, (1) but hearing loss (as with all disabilities) is more appropriately viewed as a continuum. Aging users, for example, might not identify themselves as deaf or hard of hearing, but they, too, can benefit substantially from access to alternative means for information transmission.
For the United States federal government, the need to provide accessible audio information extends beyond the market size of employees or citizens who are deaf and hard of hearing. In 2001, an amendment to Section 508 of the Rehabilitation Act mandated that federal agencies purchase only hardware, software, services, and documentation that are accessible to users with disabilities. Federal agencies have an internal mandate to ensure that all of the information they provide to the public or to their employees be accessible.
Considerable Web accessibility activity has addressed the requirements of blind and low-vision users. Failure to provide the appropriate infrastructure for such users can effectively eliminate their ability to access IT, thereby creating a profound digital divide. Less attention has been focused on the requirements of full access for users who are deaf and hard of hearing. Creating IT information in formats accessible to blind users requires mindfulness on the part of developers to create software that does not functionally lock out nonsighted users. Developers can be taught to include text tags with graphics, for example, thereby increasing accessibility for nonsighted users. Ensuring access for deaf users, on the other hand, is typically not in the hands of the developer. Deaf users can access visual IT information, but they are locked out in cases where audio information becomes prevalent. This is occurring with increasing frequency on the Web, as more and more information is presented in multimedia formats. Ensuring that audio information has associated captions or sign interpretation exceeds the bounds of the software developer's job description.
Deaf users can successfully navigate much of the educational and business space as long as they require access only to data that is presented visually. However, significant information in the workplace and in academic settings is transmitted through audio channels. In the academic environment, lecture material is typically presented orally. In the workplace environment, there is substantial information transfer through audio means, including meetings, conference calls, corporate training sessions, and presentations.
There are a number of mechanisms to address this need, but they are typically only partially and inconsistently deployed. In a mainstream university setting, classes can be interpreted through a sign interpreter or through a stenographer. (2) In the workplace setting, stenographers or sign interpreters can provide information at meetings, corporate training sessions, or other business events where audio information is the communication medium. Key barriers for these solutions, however, are cost and availability. The costs associated with sign interpretation are in the range of $50 per hour. There is also a dire shortage of skilled interpreters. The costs associated with stenography are more daunting, with service ranging from $100-$200 per hour. (3) Stenography skills are also a scarce resource, with demand outstripping supply. In large United States urban centers, one deaf student can require up to $100,000 of interpreting services in the course of an academic program. One stenographer/captionist (court reporter status) could cost over $60,000 per year.
ASR presents a potential resolution to a number of these access problems, both in the workplace and in educational settings. In reality, however, ASR has fallen short in terms of providing fully accessible environments for users who are deaf and hard of hearing. In an ideal scenario, any speaker would be recognized talking on any topic, and speech would be displayed as text for listeners who are deaf and hard of hearing. No specialized training or microphone apparatus would be required. Noisy backgrounds, multiple speakers, and bandwidth-limited phone lines would also pose no problems. Although speech recognition has advanced dramatically over the last 20 years, this "holy grail" scenario has not yet been realized. (4) Successful speech recognition applications abound, but they sidestep shortcomings by limiting applications to those within the capabilities of the technology. For example, the language model may be confined to a particular topic area such as "mutual fund transactions," or speakers may only be asked specific questions with relatively constrained and predictable replies. These application design decisions result in successful automated speech recognition applications, but they cannot resolve the wider range of problems mentioned earlier that are faced by persons who are deaf and hard of hearing.
INCREMENTAL IMPROVEMENTS FOR DEAF ACCESSIBILITY
Several approaches have been pursued toward the goal of providing incrementally better accessibility for persons who are deaf and hard of hearing. These approaches are discussed in the following sections.
Remote stenography
One of the techniques mentioned as an access tool for deaf individuals is stenographic transcription. This approach is deployed widely in corporate settings, particularly for large gatherings such as public meetings or presentations by executives. For more casual meetings, however, the problem remains. Employing a stenographer on site for routine meetings becomes prohibitively expensive, and scheduling must be done with significant lead time. To address this problem, a number of stenography companies have begun introducing remote stenography options. As a case in point, the stenography company Caption First, working with consulting assistance from IBM Research, has created a tool called netCAPTION. (5) This tool allows stenographic output to be transmitted as streaming text over the Internet, providing a number of advantages. In particular, captioning is available without requiring that the stenographer physically be on site. The stenographer participates in the remote meeting via conference call, and the deaf participant is able to view the captions in near real time over the Internet. This reduces the cost of transcription, as...
NOTE: All illustrations and photos
have been removed from this article.
More articles from IBM Systems Journal
Personalization, interaction, and navigation in rich multimedia docume..., September 01, 2005
Semantic triage for increased Web accessibility.(analysis on web site ..., September 01, 2005
Looking for additional articles?
Search our database of over 3 million articles.
Looking for more in-depth information on this industry?
Search our complete database of Industry & Market reports by text, subject, publication
name or publication date.
About Goliath
Whether you're looking for sales prospects, competitive information, company
analysis or best practices in managing your organization,
Goliath can help you meet your business needs.
Our extensive business information databases empower business
professionals with both the breadth and depth of credible,
authoritative information they need to support their business
goals. Whether it be strategic planning, sales prospecting,
company research or defining management best practices -
Goliath is your leading source for accurate information.
|
%20%20&position=1)