Towards best practice in the development and evaluation of speech recognition components of a spoken language dialog system

Abstract

This article provides a global overview of the main aspects of current practice in the design, implementation and evaluation of speech recognition components for Spoken Language Dialog Systems (SLDSs), and presents the results of the DISC European project related to speech recognition. DISC and its successor DISC-2 are efforts towards the definition of best practice guidelines for SLDS development and evaluation. SLDSs aim at using natural spoken input for performing an information processing task such as automated standards, call routing or travel planning and reservations. The main functionality of an SLDS are speech recognition, natural language understanding, dialog management, database access and interpretation, response generation and speech synthesis. Speech recognition, which transforms the acoustic signal into a string of words, is a key technology in any SLDS. † This paper is based on research carried out within the ESPRIT 4th Framework LTR Concerted action projects 24823 and 29597 DISC-Spoken Language Dialog Systems and Components; Best Practice in Development and Evaluation. 1 DISC addresses the following six aspects of SLDSs: speech recognition, speech generation, language understanding and generation, dialogue management, human factors and systems integration. See Section 2.

Requirement Engineering Contributions to Voice User Interface

“…These systems can be defined by six features [11,12,18]: • Speech Recognition: process that allows a computer to convert an acoustic sign to a text, setting the user free from the WIMP -Windows, Icons, Menus, Pointers -interface restriction. This process includes the following steps: capturing the acoustic sign; converting this sign to digital components which are, then, divided into different sounds; classifying the sound, establishing all the possible combinations among the sound divisions and the phonetic representations; and searching for a word in the vocabulary that best fits this sound combination;…”

“…Many of these questions are derived from interface usability criteria, such as the feedback. Lamel, Minker and Paroubek [12] provide an overview of how to build this type of systems, but do not particularly specify the requirements for them.…”

Exploratory analysis of real personal emergency response call conversations: considerations for personal emergency response spoken dialogue systems

“…Chen et al [27] designed an automatic speech recognition (ASR) module with limited vocabulary for use in accessing personal emergency assistance by older adults on a mobile cell phone platform. An ASR is a basic component of all SDS [24, 28–30]. Kim et al [31] proposed a multi-modal approach to PER using inputs from vision, voice, and a body worn gravity sensor.…”

Avaliação de usabilidade para sistemas de transcrição automática de laudos em radiologia.

“…O processo de compreensão da linguagem falada pode ser, por exemplo, a transformação de uma fala em texto, através do reconhecimento da fala. Este reconhecimento é o processo de extração automática da informação linguística do sinal da voz (LAMEL;PAROUBEK, 2000;LAI, 2000;SAN-SEGUNDO et al, 2005, COHEN;GIANGOLA;BALOGH, 2004;MCTEAR, 2002 Um sistema com interface do usuário baseada em voz típico é composto por módulos conforme mostrado na Figura 3 (LAMEL;PAROUBEK, 2000;LAI, 2000;SAN-SEGUNDO et al, 2005, COHEN;GIANGOLA;BALOGH, 2004;MCTEAR, 2002).…”