Results 1 - 10
of
13
Indexing and Retrieval of Broadcast News
- Speech Communication
, 2000
"... This paper describes a spoken document retrieval (SDR) system for British and North American Broadcast News. The system is based on a connectionist large vocabulary speech recognizer and a probabilistic information retrieval system. We discuss the development of a realtime Broadcast News speech r ..."
Abstract
-
Cited by 33 (7 self)
- Add to MetaCart
This paper describes a spoken document retrieval (SDR) system for British and North American Broadcast News. The system is based on a connectionist large vocabulary speech recognizer and a probabilistic information retrieval system. We discuss the development of a realtime Broadcast News speech recognizer, and its integration into an SDR system. Two advances were made for this task: automatic segmentation and statistical query expansion using a secondary corpus. Precision and recall results using the Text Retrieval Conference (TREC) SDR evaluation infrastructure are reported throughout the paper, and we discuss the application of these developments to a large scale SDR task based on an archive of British English broadcast news. Keywords: Spoken Document Retrieval; Information Retrieval; Broadcast Speech; Large Vocabulary Speech Recognition. 1 Introduction Retrieval of audio segments according to their content is a challenging and significant problem. It has been estimated th...
Discriminating Capabilities of Syllable-based Features and Approaches of Utilizing Them for Voice
- Retrieval of Speech Information in Mandarin Chinese,” IEEE Trans. on Speech and Audio Processing
, 2002
"... Abstract—With the rapidly growing use of the audio and multimedia information over the Internet, the technology for retrieving speech information using voice queries is becoming more and more important. In this paper, considering the monosyllabic structure of the Chinese language, a whole class of s ..."
Abstract
-
Cited by 31 (14 self)
- Add to MetaCart
Abstract—With the rapidly growing use of the audio and multimedia information over the Internet, the technology for retrieving speech information using voice queries is becoming more and more important. In this paper, considering the monosyllabic structure of the Chinese language, a whole class of syllable-based indexing features, including overlapping segments of syllables and syllable pairs separated by a few syllables, is extensively investigated based on a Mandarin broadcast news database. The strong discriminating capabilities of such syllable-based features were verified by comparing with the word- or character-based features. Good approaches for better utilizing such capabilities, including fusion with the word- and character-level information and improved approaches to obtain better syllable-based features and query expressions, were extensively investigated. Very encouraging experimental results were obtained. Index Terms—Confidence measure, retrieval of speech information, syllable-based features, term association matrix. I.
Effects of word recognition errors in spoken query processing
- Proceedings of the IEEE ADL 2000 Conference (pp. 39–47). Washington DC
, 2000
"... The effects of word recognition errors (WRE) in Spoken Document Retrieval have been well studied and well re-ported in recent Information Retrieval (IR) literature. Much less experimental work has been devoted to studying the ef-fects of WRE in Spoken Query Processing in IR. It is easy to hypothesiz ..."
Abstract
-
Cited by 9 (8 self)
- Add to MetaCart
(Show Context)
The effects of word recognition errors (WRE) in Spoken Document Retrieval have been well studied and well re-ported in recent Information Retrieval (IR) literature. Much less experimental work has been devoted to studying the ef-fects of WRE in Spoken Query Processing in IR. It is easy to hypothesize that given the typical length of the user query, the effects of WRE in spoken queries on the performance of IR systems must be destructive. The experimental work re-ported in this paper intends to test that. The paper reports on the background of such a study, on the construction of a suitable test collection, on the first experimental results ob-tained and on the limitations of the study. The results show that classical IR techniques are quite robust to considerably high levels of WRE rates in spoken queries (roughly below 40%), in particular for long queries. 1
Hybrid Approach to Spoken Query Processing in Document Retrieval System
- In Proc. ESCA Workshop on Accessing Information In Spoken Audio
, 1999
"... In the context of the THISL spoken document retrieval system, we present a hybrid approach to spoken query processing, which enables to increase recognition rates and to extract relevant information for the application. The query processing is distributed between grammar and language model, based on ..."
Abstract
-
Cited by 4 (0 self)
- Add to MetaCart
(Show Context)
In the context of the THISL spoken document retrieval system, we present a hybrid approach to spoken query processing, which enables to increase recognition rates and to extract relevant information for the application. The query processing is distributed between grammar and language model, based on the assumption that a query can be decomposed in two relatively independent parts; the addressing form, which is parsed with a grammar, and the queried content, which is scored with a domain specific language model. Our aim is to retrieve the content sequence, which allows us to consult the database, but also, to keep information about the query formulations in order to develop an interaction between the user and the retrieval engine. This leads us to work closely with the speech recogniser and to carry out together the recognition and the query analysis. 1. Introduction Speech recognition technology has now reached a stage where it can reasonably provide baseline systems for spoken inter...
Combination of similarity measures for effective spoken document retrieval
- Journal of Information Science
"... Often users of information retrieval systems and document authors use different terms to refer to the same concept. For this simple reason, information retrieval is affected by the ‘term mismatch ’ problem. The term mismatch problem does not only have the effect of hindering the retrieval of relevan ..."
Abstract
-
Cited by 4 (0 self)
- Add to MetaCart
Often users of information retrieval systems and document authors use different terms to refer to the same concept. For this simple reason, information retrieval is affected by the ‘term mismatch ’ problem. The term mismatch problem does not only have the effect of hindering the retrieval of relevant documents, it also produces bad rankings of relevant documents. A similar problem can be found in spoken document retrieval, where terms misrecognized by the speech recognition process can hinder the retrieval of potentially relevant spoken documents. We will call this problem ‘term misrecognition’, by analogy to the term mismatch problem. This paper presents two classes of retrieval models that attempt to tackle both the term mismatch and the term misrecognition problems at retrieval time using term similarity information. The models use either complete or partial knowledge of semantic and phonetic term similarity, evaluated using statistical methods from the corpus. 1.
Grammar and N-gram Collaboration for Information Retrieval Interface
- In SPECOM'98, ESCA---ELSNET International Workshop on Speech and Computer, St
, 1998
"... In the context of a spoken document retrieval system, we present an algorithm for searching word lattices, which enables to increase recognition rates and to extract relevant information for the application. The effort is distributed between grammar and language model, based on the assumption that a ..."
Abstract
-
Cited by 2 (2 self)
- Add to MetaCart
(Show Context)
In the context of a spoken document retrieval system, we present an algorithm for searching word lattices, which enables to increase recognition rates and to extract relevant information for the application. The effort is distributed between grammar and language model, based on the assumption that a query can be decomposed in two relatively independent parts; the addressing form, which is parsed with a grammar, and the queried content, which is scored with a domain specific language model. 1. Introduction Speech recognition technology has reached a stage where it can reasonably provide baseline systems for spoken interfaces, especially for semantically-light applications, like call-attendance or keyword based document retrieval. Yet, the design of vocal interface must address the problem of recognition errors, utterance disfluencies and lack of cover of the interpretation process. 1.1. What recognition output? As far as recognition errors are concerned, most experimentations emphasi...
An Experimental Study of the Effects of Word Recognition Errors in Spoken Queries on the Effectiveness of an Information Retrieval System
, 1999
"... The effects of word recognition errors (WRE) in spoken documents on the performance of an Information Retrieval (IR) system have been well studied and well reported in recent IR literature. Most of the research in this direction has been promoted by the Spoken Document Retrieval track of TREC. Much ..."
Abstract
-
Cited by 1 (0 self)
- Add to MetaCart
(Show Context)
The effects of word recognition errors (WRE) in spoken documents on the performance of an Information Retrieval (IR) system have been well studied and well reported in recent IR literature. Most of the research in this direction has been promoted by the Spoken Document Retrieval track of TREC. Much less experimental work has been devoted to studying the effects of WRE in spoken queries. It is easy to imagine that given the typical length of the user query, the effects of WRE in queries on the performance of an IR system must be destructive. The experimental work reported in this paper intends to test that. The paper reports on the background of such a study, on the construction of a test collection, and on the first experimental results. The preliminary conclusions drawn from the experimentation enable to give some useful indications for the design of spoken query systems, despite the recognized limitations of the study. International Computer Science Institute, 1947 Center St. Suite ...
c ○ 2000 Kluwer Academic Publishers. Manufactured in The Netherlands. New Approaches to Spoken Document Retrieval
, 1999
"... Abstract. This paper presents four novel techniques for open-vocabulary spoken document retrieval: a method to detect slots that possibly contain a query feature; a method to estimate occurrence probabilities; a technique that we call collection-wide probability re-estimation and a weighting scheme ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract. This paper presents four novel techniques for open-vocabulary spoken document retrieval: a method to detect slots that possibly contain a query feature; a method to estimate occurrence probabilities; a technique that we call collection-wide probability re-estimation and a weighting scheme which takes advantage of the fact that long query features are detected more reliably. These four techniques have been evaluated using the TREC-6 spoken document retrieval test collection to determine the improvements in retrieval effectiveness with respect to a baseline retrieval method. Results show that the retrieval effectiveness can be improved considerably despite the large number of speech recognition errors. Keywords: spoken document retrieval, speech recognition, retrieval effectiveness 1.
ISCA Archive A HYBRID APPROACH TO SPOKEN QUERY PROCESSING IN DOCUMENT RETRIEVAL SYSTEM ✶
"... In the context of the THISL spoken document retrieval system, we present a hybrid approach to spoken query processing, which enables to increase recognition rates and to extract relevant information for the application. The query processing is distributed between grammar and language model, based on ..."
Abstract
- Add to MetaCart
(Show Context)
In the context of the THISL spoken document retrieval system, we present a hybrid approach to spoken query processing, which enables to increase recognition rates and to extract relevant information for the application. The query processing is distributed between grammar and language model, based on the assumption that a query can be decomposed in two relatively independent parts; the addressing form, which is parsed with a grammar, and the queried content, which is scored with a domain specific language model. Our aim is to retrieve the content sequence, which allows us to consult the database, but also, to keep information about the query formulations in order to develop an interaction between the user and the retrieval engine. This leads us to work closely with the speech recogniser and to carry out together the recognition and the query analysis. 1.
