Takayuki Watanabe

文科省の管理下にある順応型組織の大学で生じている様々な教育課題を解決するために、ブロックチェーンの技術を使って大学を多元型・進化型組織の「Web3大学」に進化させ、自律分散化した大学教員コミュニティの自由な競争と、大学の壁を超えた自由な教育の提供と履修により教育を向上させることを目指しています。

We made the audio description narrated with TTS and compared it with that with human voice. We found that (1) the guide with TTS obtained the same appreciation if the voice and mood of TTS was matched with an atmosphere of a movie, (2) evaluation of the guide varied depending on liking for TTS voice by users, and (3) there are two types of the visually impaired who want art for a guide and who want information transmission efficiency for it. In addition to that, production time of a guide must be shortened if a guide is made with TTS. We, therefore, concluded that if audio description is made with TTS, it must be easier for everyone to make audio description.

The objective of the current research is to investigate if the Speacon, which is created by speeding up a TTS so that it is no longer comprehensible as a particular word, is useful to auditory represent the logical structure of Web (e.g. heading elements). We have carried out an experiment with g blind subjects with ordinal screen-reader and the specially arranged screen-reader which uses Speacon to represent Web structure. No statistical differences were found in terms of task completion rate, usability, and mental-workload. Most subjects. however, favored Speacon because it was easy to distinguish structure information and content and it took shorter time to represent structural information. We also found that we must pay attention to practice effect, easy-to-hear TTS when spoken fast, and appropriate setting of voice and pitch for each subject.

The objective of this article is to describe how and what Japan has carried out standard harmonization, both internationally with WCAG 2.0 and nationally in Japan, and achievement efforts of Web accessibility based on the harmonized national Web content accessibility standard, JIS X 8341-3: 2010. In this article, I firstly explain the importance of Web accessibility for the disabled and the aged, and show national standard series of JIS X 8341 "Guidelines for older persons and persons with disabilities -- Information and communications equipment, software and services." Then, the development and revision of JIS X 8341-3 is explained from a viewpoint of harmonization effort with WCAG 2.0. Thirdly, Japanese activities to improve Web accessibility based on JIS X 8341-3: 2010 are explained in detail. In conclusion, Japan has a harmonized national industrial standard of Web content accessibility but there are many things to tackle to improve Web accessibility.

The ability of information search is needed to use the Internet efficiently and effectively. In this research, we conducted some investigations and experiments with the particular three older people to show the purpose and the characteristics of their web usage. As a result, we found that old people used the Internet as a way to fulfill their needs of a real life. In terms of their characteristics of web usage, we also found that they watched at the incorrect web pages again and again, they had poor ability to enter the appropriate keywords, and they needed to extra time to judge whether the web page had information they search. Therefore, it could be clear that some kind of navigation could help them use the Internet efficiently and that web site designer or developer must pay attention to the characteristics of old people group such as a various age, PC skills, and their web experiences.

This research investigates what kind of auditory representation is suited for logical structure and visual representation of Web content. We carried out experiments with 24 female university students. We showed them various representations and measured correctness and mental workload. Experimental results show that Spearcons is most suited to represent logical structures such as heading elements, and Spearcons and sign-sound is suited for visual representations such as bold font or large fonts. Sign-sound is more suited than Spearcons to represent visual representations because there are many confusing words to represent visual presentations (in Japanese).

This research focused on "Audio description of a movie." Audio description is a narration explaining visual information. We were interested in the idea, what kind of audio description well expresses the 3D-effect of a three-dimensional movie? We found that the visually impaired didn't want verbalizing the 3D-effect. Next, we carried out an experiment which compared normal audio description and 3D audio description with visually-impaired subjects. As a result, we found that although they generally preferred normal audio description, some preferred 3D audio description at the following cases: 1) those who think story is important for a movie think 3D audio description is suitable for mystery movies, 2) those attracted by sounds think 3D audio description is suitable for action movies.

It is important for Web usability that the mental model drawn by a user is same as that by a developer. However, it is difficult for a developer to construct an appropriate Web content with that model. We carried out experiments that used three kinds of Web sites to clarify mental model gaps between the elderly users and developers. The results show that the site, which had obvious navigation and the simplest design, is the most usable. We also find that there are some functions which were implemented by the developers as important functions but are not used by the users.

Speech-based CAPTCHA systems, which distinguish between software agents and human beings, are especially important for persons with visual disability. The popular approach is based on mixing-based methods, which use the mixed sounds of target speech and noises. We have proposed a deletion-based method which uses the phonemic restoration effects. Our approach can control the difficulty of tasks simply by the masking ratio. Our design principle of CAPTCHA insists that such tasks should be chosen so that the larger difference in performance between the machines and human beings can be provided. In this paper, we give some hypotheses on the differences between the deletion-based method and the mixing-based methods. We also show a plan of experiments which compare the automatic speech recognition performance, speech intelligibility, and mental workload of these two approaches.

We investigated the intelligibility of ultra-fast speech which may be used for screen reader for persons with visual disability. The subjects were divided into four groups and they listened to 150 words with the speed of approximately 20 morae/sec. The vocabulary contained the tasks of high and low familiarity words, and the orders of tasks were different by the groups. Four morae Japanese words from the FW03 database were used as the vocabulary of the recall test. As a result, significant learning effect was observed in cases where the subject listened the high familiarity words in succession. This indicates that the learning effect to the ultra-fast speech is promoted when the mental lexical access is easy.

We carried out preliminary experiments to realize a speech-based CAPTCHA system to distinguish between software agents and human beings. Such systems are especially important for persons with visual disability. As the first step, we investigated the design principles of CAPTCHA systems from viewpoint of creating the performance gap between the machines and human beings. Based on hypothesis that the Phonemic Restoration effects are useful for this purpose, we carried out experiments of intelligibility tests by human beings and speech recognition tests with HMM-based systems.

The number of older persons (whose age is over 65) is increasing in Japan. Internet usage by older persons, however, remains low. Thus, usability and accessibility of Web content for older persons is important to increase their usage. We carried out an experiment which used two kinds of Web sites, the site A which consisted of one page and the site B which consisted of several pages but had the same information as the site A. User testing with 11 tasks and a questionary and an interview was carried out both for 12 older persons and 12 young female university students. Task completion time was longer for older persons than for young persons. It was shorter for the site A for some tasks but shorter for the site B for another tasks. The site B gave higher satisfaction for both older and young persons. These results show that the site B is more usable for older persons as well as young persons.

We investigated the intelligibility of ultra-fast speech which may be used for screen reader for persons with visual disability. The subjects were 35 women who are university students and are not visually/hearing impaired. They were divided into four groups and they listened to 150 words with the speed of approximately 20 morae/sec. The vocabulary contained the tasks of high and low familiarity words, and the orders of tasks were different by the groups. Four morae Japanese words from the FW03 database were used as the vocabulary of the recall test. As the results, we observed the significant learning effect. The transfer of the learning effects among the different word familiarities was also observed. We also investigated the mental workloads of the listening task using the NASA-TLX method. As the results, significantly high workload scores were observed at the listening of low familiarity words. However, there were great differences of workload scores between individuals. No significant reduction of workload scores was observed as the effect of learning.

Rich Internet Applications, which interact with users dynamically and which content changes dynamically, are emerged with Ajax and JavaScript technologies. These dynamic Web pages, however, may cause accessibility problems. We carried out a case study with two Google applications, iGoogle and Google document, to examine 1) what kind of accessibility problems are came out, and 2) which problems are resolved by a code with good practice and which problems is inevitable.

We investigated the intelligibility of ultra-fast speech which may be used for screen reader for persons with visual disability. The subjects were 35 women who are university students and are not visually/hearing impaired. They were divided into four groups and they listened to 150 words with the speed of approximately 20 morae/sec. The vocabulary contained the tasks of high and low familiarity words, and the orders of tasks were different by the groups. Four morae Japanese words from the FW03 database were used as the vocabulary of the recall test. As the results, we observed the significant learning effect. The transfer of the learning effects among the different word familiarities was also observed. We also investigated the mental workloads of the listening task using the NASA-TLX method. As the results, significantly high workload scores were observed at the listening of low familiarity words. However, there were great differences of workload scores between individuals. No significant reduction of workload scores was observed as the effect of learning.

An audio description is important for the visually impaired. We have developed a beginner's manual for a live description, which describes a movie showed in a movie theater. In order to make an objective and trust-worthy manual, we carried out interviews with novice and skillful describers as well as visually impaired audience, analysis described content of skillful describers, and carried out an experiment to examine what part of the movie is described by every describer.

Increasing Web usage requires Web accessibility, use of Web including people with disabilities and older people. W3C, which is known as a de facto Web standard developer, is preparing WCAG 2.0, a Web content accessibility guidelines. WCAG 2.0 is expected to become W3C Recommendation in 2007 and will be a worldwide accessibility standard. This paper describes: 1) the outline of the current WCAG 2.0 working draft, and 2) how to improve Web accessibility with use of guidelines.

A prototype lunch delivery Web system for the visually impaired was developed based on the analysis of human (the visually impaired who wants to order a lunch) to human (the sighted who helps the visually impaired to decide lunch by reading aloud lunch menu) dialog. Based on these analysis, a prototype system was developed, which consists of three steps : 1) rough selection (candidate items are roughly selected based on categories), 2) selection of favorites (favorite items are selected from candidate items and stored in the system), and 3) final selection (one item is selected with detailed information for each item). To generate XHTML document dynamically, PHP and PostgreSQL were used. The evaluation of the prototype system is performed by the visually impaired only with synthesized speech and key-board. The results showed that our system is effective and easy to use.

Many web developers tend to depend on automated test tools to improve web accessibility. It however is important to observe real users who use assistive technologies to interact with web. In order to understand web accessibility problems correctly, user-centered analysis is indispensable. We, as a baseline study, carried out user testing with blind users and hold an interview with an instructor who had taught how to use web for disabled people. We found that (1) users rarely used dedicated functions for reading tables and headings but just read content line by line, and (2) there was a content which was passed by a test tool but didn't to users' navigation strategies.

インターネットの普及とWebの利用範囲の拡大とともに，高齢者や障害者を含む誰もがWebを使えることの重要性が高まっている。Webの標準仕様を策定している国際コンソーシアムW3Cは，このようなアクセシブルなWebを制作するためのガイドラインであるWeb Content Accessibility Guidelinesの改訂版（WCAG 2.0）を策定中である。WCAG 2.0は，順調にいけば2007年度中にW3Cの正式勧告となり，世界中に影響を及ぼすと思われる。本論では，まずWCAG 2.0ワーキングドラフトを解説し，次に，WCAGなどのガイドラインを活用してアクセシブルなWebを推進するために必要な基礎知識を述べる。

We investigated the intelligibility of ultra-fast speech which may be used for screen reader for persons with visual disability. The subjects were 35 women who are university students and are not visually/hearing impaired. They were divided into four groups and they listened to 150 words with the speed of approximately 20 morae/sec. The vocabulary contained the tasks of high and low familiarity words, and the orders of tasks were different by the groups. Four morae Japanese words from the FW03 database were used as the vocabulary of the recall test. As the results, we observed the significant learning effect. The transfer of the learning effects among the different word familiarities was also observed. We also investigated the mental workloads of the listening task using the NASA-TLX method. As the results, significantly high workload scores were observed at the listening of low familiarity words. However, there were great differences of workload scores between individuals. No significant reduction of workload scores was observed as the effect of learning.