IBM researcher Chieko Asakawa has been blind since she was 14 years old. Since joining IBM in Japan in 1985, she has worked on myriad projects to improve accessibility for the visually impaired. Asakawa, now a senior accessibility researcher at IBM's Tokyo Research Laboratory, has been working on nonvisual computer interfaces in an effort to improve Web accessibility and usability for the visually impaired and others with special needs. She helped develop the IBM Home Page Reader in 1997 and a digital Braille system and three key applications, including a Braille Editing System to allow users to easily input and edit Braille using an ordinary keyboard and monitor. In 2004, she and her team also previewed a disability simulator that helps Web designers ensure that their pages are accessible and usable by visually impaired users. And over the last year, she and a team of four researchers have been working to make it possible for blind and visually impaired users to access multimedia content online, using a keyboard to control media player software. She discussed her work last week via e-mail from Tokyo.
Excerpts from that interview follow:
As a visually impaired researcher and active Internet user, was this project motivated by your personal situation -- knowing that all kinds of Web content was out there that you and other visually impaired people couldn't access?
Increasingly, I have been facing difficulties where I simply could not access Web content easily. I recently conducted a survey and found that most of the tested Web sites with multimedia content were not accessible. Based on this result and my personal experience, I fear that if we don't take any action, it will broaden the digital gap between the sighted and the blind. Today, accessibility for static HTML has been well established in various aspects, including technical, guidelines and regulations. And indeed, static Web contents have helped us narrow the gap. I felt that there should be ways to help narrow the gap concerning multimedia content's accessibility.
The tool set provides the user with keyboard-controlled ways to run some media player applications, such as starting the video, stopping it, rewinding it, etc.?
Without the tool set, these functions can't be controlled with a keyboard? The tool is compatible with Windows Media Player and Flash. Users only need to know a unified shortcut key operation to run video and animation. Previously, these functions could not be controlled by using a keyboard since the images that are up on the Web sites are only controllable by pointing and clicking a mouse, especially for embedded players in Web pages. Only very rarely, there might be a case where there are play/stop and volume up/down buttons that can be operated with a keyboard. However, it is hardly possible to find out the existence of such buttons while video is playing due to conflicts with screen reading software.
Once you identified the problem, how did you come up with your ideas for accessing streaming media, video and other visual content online?
First, I thought if we can provide a function to separately control volume, at least, we will be able to hear multimedia sound and screen reader sound. Technically, it was not easy to provide such a function. However, given the inability to access multimedia content pages, we did not give up; we thought we could find the way. Second, it's really frustrating to wait until inaccessible pages become accessible [by page designers who know to prepare content for visually impaired users]. If we can make inaccessible pages accessible by providing external metadata, we can significantly shorten the time to access such pages. We developed a tool [that] analyzes and adapts external metadata dynamically, and it can generate accessible pages on the fly on the client side.
Can you describe the tool set that you developed? What computer language is it written in? How does it work? How does it interact with other programs such as media players?
The tool is mostly written in Java, and works as a standalone application for users. When a user starts the application, it gives the same experience as when you open Internet Explorer and surf any Web pages. Once a user opens a Web page, the browser automatically analyzes multimedia objects inside the page, then the browser [establishes] a connection to each multimedia object. Currently, it has adaptors for [Adobe] Flash and Windows Media Player. A part of the adaptors are written in C++.
Usually, multimedia content is only designed for mouse operation, so it is impossible or too hard to operate by using a keyboard. The tool has a function to provide an alternative text-based interface to the content based on manually created XML metadata. Someone needs to create the metadata manually, but once created, usability of the site is drastically improved. It also provides functions to add audio descriptions to movies based on XML metadata. Audio descriptions are usually created by content creators as an additional sound track of a movie. While descriptions should be authored manually in advance, our tool provides a cost effective way for anyone who wants to create [and add] the audio description [later] to help blind users. Without making any changes to the content itself, the attached XML metadata information can synchronize with the video.
Lastly, the tool has a function [that] controls speech rate. Visually impaired users are well accustomed to high-speed voice, since we are using screen readers [that read text aloud for the user] every day. As for videos, it is better to provide speed control functions, but it was technically difficult. So, we developed some technology to control speech rate of any types of players, including Flash, Media Player and any types of media content can be sped up by using the tool.