Interaction Design Group

Following up with the speech technology development, spoken dialogue systems are getting commercialized. Many public transportation companies, call centres, tourist information centres, etc are using spoken dialogue systems instead of the touch-tone teleco

User’s Experience of a Commercial Speech Dialogue System

Fang Chen, Yael Katzenellenbogen

Interaction Design Group

Department of Computer Science and Engineering Chalmers University of Technology, Gothenburg, Sweden

fanch@cs.chalmers.se

Abstract Following up with the speech technology development,

spoken dialogue systems are getting commercialized. Many

public transportation companies, call centres, tourist

information centres, etc are using spoken dialogue systems instead of the touch-tone telecommunication system for

customer communication. However, there are many usability

problems with the public application of spoken dialogue systems. An evaluation of a Swedish dialog system is

conducted through a test with six users who went through

three different scenarios in a home environment. The results

came from analysis of the subjects’ conversation with the

dialogue system, and from observations, interviews and

questionnaires. The result shows that the system is functional,

since all the users concluded their tasks. However, there are a

few usability problems that need improvement, such as the

understanding between the user and the system, the systems

flexibility, naturalness, and correctness, as well as the amount

of information, error handling and the users control of the

system. Concerning the users’ experience, the study showed

that variables such as elegance, emotion, encouragement,

intelligence or personality were not present. These variables

would affect users’ attitude towards the system.

1. Introduction Nowadays, when you pick up a telephone and call a company,

you don’t often get connected to the human operator directly.

It normally takes a couple of navigation steps before you can

talk to the operator, and sometimes you may have to wait in a

long queue.

Outline the recent progress of speech technology on the

communication system, automatic information services via

telephone is becoming one of the most important application

areas of speech technology, and has been rapidly developing.

Applications via telecommunication and various telephone

voice services such as finance, call center and supply chain,

tourist information access, ticket booking for airlines, buses,

and railways, order-entry, school or medical information

access, etc are more or less working in the real context, or

under the investigation.

Compare with the touch-tone interactive voice

recognition systems, automatic information access via speech

provides the most effectively and enterprises

telecommunications service. Touch-tone voice recognition

systems are easy to perform but have a very limited flexibility

and capacity. The usage alienates many customers due to

tedious and frustrating push-button interfaces. Spoken

dialogue systems, on the other hand, can overcome the

problems in touch-tone service systems and therefore enable

companies to dramatically save costs [1, 2]

Cameron [3] analysed the success and failure of a large

number of commercial speech systems developed in the U.S.

over the last decade and concluded that people will use speech when: They are offered no choice It corresponds to the privacy of their surroundings Their hands or eyes are busy on another task It is quicker than any alternative. The first three reasons relate in varying degree to external

constraints on the user. The last one relates to the speech

system designer’s expectation, but it is rarely used.

Like many other public service systems, the user

population for the public dialogue system is very large;

almost every civilian in the society shall be counted as the

potential user. To design a system that can be of universal

access, a few requirements need to be achieved 1) equitable

use; 2) flexibility in use; 3) simple and intuitive to use; 4)

perceptible information; 5) tolerance for error and 6) low

physical effort [4].

The big user population brings complexity and diversity

of the user demand, skills or experience, preference etc. The system must accept untrained, unskilled utterances that

implicitly contain unexpected forms of speech e.g. vague,

breadth, too laud or too small, distortion, some other noises, and so forth. Therefore, it must be paid much attention on dialogue system design that liberates the end users from mental stress together with trustable performance, to satisfy the usability requirement. Many dialogue system designers have put much effort in the design to fit the novice users, but have ignored the skilled users. When human talks to each other, they will soon learn about each other’s characteristics during the conversation. It is the same for human being talking to the spoken dialogue system. After the users try a couple of sentences in a conversation with the system, they will soon get an understanding of how the system is working, and become a “skilled” user. When the user communicates with the system by voice, they often expect the counter partner to be like a human, even though they know it is a machine on the other side of the telephone line [5]. They will expect the system to “learn” something about them and not treat them as the novice user any longer. How to design the system so that both a novice user, as well as a skilled user can feel be respected when talking to the dialogue system has become a new challenge to the designer. There are very few reports one can find in the literature about usability evaluation and users satisfaction of these automatic voice service systems. Dybkjear and her co-authors

搜索“diyifanwen.net”或“第一范文网”即可找到本站免费阅读全部范文。收藏本站方便下次阅读，第一范文网，提供最新人文社科Interaction Design Group全文阅读和word下载服务。