Abstract — One significant problem for spoken language systems is how to cope with users ’ OOD (out-of-domain) utterances which cannot be handled by the back-end application system. In this paper, we propose a novel OOD detection framework, which makes use of the classification confidence scores of multiple topics and applies a linear discriminant model to perform in-domain verification. The verification model is trained using a combination of deleted interpolation of the in-domain data and minimumclassification-error training, and does not require actual OOD data during the training process, thus realizing high portability. When applied to the “phrasebook ” system, a single utterance read-style speech task, the proposed approach achieves an absolute reduction in OOD detection errors of up to 8.1 points (40% relative) compared to a baseline method based on the maximum topic classification score. Furthermore, the proposed approach realizes comparable performance to an equivalent system trained on both in-domain and OOD data, while requiring no OOD data during training. We also apply this framework to the “machineaided-dialogue” corpus, a spontaneous dialogue speech task, and extend the framework in two manners. First, we introduce topic clustering which enables reliable topic confidence scores to be generated even for indistinct utterances, and second, we implement methods to effectively incorporate dialogue context. Integration of these two methods into the proposed framework significantly improves OOD detection performance, achieving a further reduction in EER of 7.9 points. I.

We present an implemented concept-to-speech (CTS) systh tl/at offers original proposals for certain couplings-of dialogue computation with prosedie computation. Specifically, the semantic interpretation, task modeling and dialogue strategy n,odules in a working spoken dialogue system are used to generate prmodic features to better convey the meaning of system replies. The new GTS system embodies and extends theoretical work on intona- tional meaning in a more general, robust and rigorous way than earlier approaches, by reflecting compositional aspects of both dialogue and intonation intereprctation in an original computational framework for prosedie generation.

In this paper we describe a system that uses natural language dialogue to provide customer service for a financial domain both over the phone as well as over the Internet. Some sample transactions include checking one's account balance, transferring money from one account to another, buying, selling, and trading stocks, etc. The system described is robust enough to handle speech recognition error rates as high as 50%. In addition, the architecture of the system has been designed such that the domain-dependent information is abstracted away into self-contained modules thus allowing for ease of portability of the system to new application domains. Introduction Currently people deal with customer service centers either over the phone or on the world wide web on a regular basis. These service centers support a wide variety of tasks including checking the balance of a bank or a credit card account, transferring money from one account to another, buying airline tickets, and filing one's in...

We describe an approach("variant transduction") aimed at reducing the effort and skill involved in building spoken language interfaces.

This paper computationalizes two linguistic concepts, contrast and focus, for the extraction of pragmatic and semantic salience from spontaneous speech. Contrast and focus have been widely investigated in modern linguistics, as categories that link intonation and information/discourse structure. This paper demonstrates the automatic tagging of contrast and focus for the purpose of robust spontaneous speech understanding in a tutorial dialogue system. In particular, we propose two new transcription tasks, and demonstrate automatic replication of human labels in both tasks. First, we define focus kernel to represent those words that contain novel information neither presupposed by the interlocutor nor contained in the precedent words of the utterance. We propose detecting the focus kernel based on a word dissimilarity measure, part-of-speech tagging, and prosodic measurements including duration, pitch, energy, and our proposed spectral balance cepstral coefficients. In order to measure the word dissimilarity, we test a linear combination of ontological and statistical dissimilarity measures previously published in the computational linguistics literature. Second, we propose identifying symmetric contrast, which consists of a set of words that are parallel or symmetric in linguistic structure but distinct or contrastive in meaning. The symmetric contrast identification is performed in a way similar to the focus kernel detection. The effectiveness of the proposed extraction of symmetric contrast and focus kernel has been tested on a Wizard-of-Oz corpus collected in the tutoring dialogue scenario. The corpus consists of 630 non-single word/phrase utterances, containing approximately 5700 words and 48 minutes of speech. The tests used

The goal of this paper is to give a description of the state of the art, the issues, the problems, and the solutions related to industrial dialog systems for the automation of technical support. After a general description of the evolution of the spoken dialog industry, and the challenges in the development of technical support applications, we will discuss two specific problems through a series of experimental results. The first problem is the identification of the call reason, or symptom, from loosely constrained user utterances. The second is the use of data for the experimental optimization of the Voice User Interface (VUI). 1

Abstract. We present a novel approach to developing interfaces for multi-application dialogue systems. The targeted interfaces allow transparent switching between a large number of applications within one system. The approach, based on the Rapid Dialogue Prototyping Methodology (RDPM) and the Vector Space Model techniques, is composed of three main steps: (1) producing finalized dialogue models for applications using the RDPM, (2) designing an application interaction hierarchy, and (3) navigating between the applications based on the user’s application of interest. 1

This paper presents a novel technique that allows testing spoken dialogue systems by means of an automatic generation of conversations. The technique permits to easily test spoken dialogue systems under a variety of lab-simulated conditions, as it is easy to vary or change the utterance corpus used to check the performance of the system. The technique is based on the use of a module called user simulator whose purpose is to behave as real users when they interact with dialogue systems. The behaviour of the simulator is decided by means of diverse scenarios that represent the goals of the users. The simulator aim is to achieve the goals set in the scenarios during the interaction with the dialogue system. We have applied the technique to test a dialogue system developed in our lab. The test has been carried out considering different levels of white and babble noise as well as a VTS noise compensation technique. The results prove that the dialogue system performance is worse under the babble noise conditions. The VTS technique has been effective when dealing with noisy utterances and has lead to better experimental results, particularly for the white noise. The technique has permitted to detect problems in the dialogue strategies employed to handle confirmation turns and recognition errors, suggesting that these strategies must be improved. q 2002 Elsevier Science B.V. All rights reserved.

In this paper, we describe the evaluation of a limited domain question-answering characters, particularly as to the effect of non-optimal speech recognition, and the ability to appropriately answer novel questions. Results show that answering ability is robust until speech recognition reaches over 60 % Word error rate. 1.

with Information and Services) has been established under joint funding from the European Commission’s 5 th Framework Program and the U.S.