Human Language Technology: Opportunities and Challenges
Published 4 October 2006
Mari Ostendorf, E. Shriberg, Andreas Stolcke
Citations21
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
This paper outlines some of the issues involved, as well as the opportunities, presenting an overview of the special double session on speech and language technologies in spoken document processing applications.
Abstract
In recent years, there has been dramatic progress in both speech and language processing, in many cases leveraging some of the same underlying methods. This progress and the growing technical ties motivate efforts to combine speech and language technologies in spoken document processing applications. This paper outlines some of the issues involved, as well as the opportunities, presenting an overview of the special double session on this topic.
Keywords
Computer Science
Combining labeled and unlabeled data with co-training
5,604 Citations1998Avrim Blum, Tom M. Mitchell
Cambridge University Press eBooksVariation across Speech and Writing
5,128 Citations1988Douglas Biber
In Variation Across Speech and Writing, six dimensions of variation are identified through a factor analysis, on the basis of linguistic co-occurence patterns, and the resulting model of variation provides for the description of the distinctive linguistic characteristic of any spoken or written text and enables reconciliation of the contradictory conclusions reached in previous research.
IEEE Signal Processing MagazineEmotion recognition in human-computer interaction
2,507 Citations2001Roddy Cowie, Ellen Douglas‐Cowie +5 more
Basic issues in signal processing and analysis techniques for consolidating psychological and linguistic analyses of emotion are examined, motivated by the PKYSTA project, which aims to develop a hybrid system capable of using information from faces and voices to recognize people's emotions.
CognitionMonitoring and self-repair in speech
1,640 Citations1983Willem J. M. Levelt
It was finally shown that the editing term plus the first word of the repair proper almost always contain sufficient information for the listener to decide how the repair should be related to the original utterance.
Computer Speech & LanguageFinding consensus in speech recognition: word error minimization and other applications of confusion networks
641 Citations2000Lidia Mangu, Eric Brill +1 more
A new framework for distilling information from word lattices is described to improve the accuracy of the speech recognition output and obtain a more perspicuous representation of a set of alternative hypotheses.
IEEE International Conference on Acoustics Speech and Signal ProcessingMinimum Phone Error and I-smoothing for improved discriminative training
629 Citations2002Daniel Povey, Philip C. Woodland
The Minimum Phone Error (MPE) and Minimum Word Error (MWE) criteria are smoothed approximations to the phone or word error rate respectively and I-smoothing which is a novel technique for smoothing discriminative training criteria using statistics for maximum likelihood estimation (MLE).
Discriminative Reranking for Natural Language Parsing
443 Citations2000Michael J. Collins
Computational LinguisticsDiscriminative Reranking for Natural Language Parsing
407 Citations2005Michael J. Collins, Terry Koo
The boosting approach to ranking problems described in Freund et al. (1998) is applied to parsing the Wall Street Journal treebank, and it is argued that the method is an appealing alternative-in terms of both simplicity and efficiency-to work on feature selection methods within log-linear (maximum-entropy) models.
Discourse segmentation of multi-party conversation
351 Citations2003Michel Galley, Kathleen McKeown +2 more
A domain-independent topic segmentation algorithm for multi-party speech that combines knowledge about content using a text-based algorithm as a feature and about form using linguistic and acoustic cues about topic shifts extracted from speech.
fMPE: Discriminatively Trained Features for Speech Recognition
278 Citations2006Daniel Povey, Brian Kingsbury +4 more
Speech repairs, intonational phrases, and discourse markers: modeling speakers' utterances in spoken dialogue
190 Citations1999Peter A. Heeman, James F. Allen
A statistical language model is presented in which the speech recognition problem is redefined so that it includes the identification of POS tags, discourse markers, speech repairs, and intonational phrases.
Automatic Dialog Act Segmentation and Classification in Multiparty Meetings
180 Citations2006Jeremy Ang, Yang Liu +1 more
It is found that a very simple prosodic model aids performance over lexical information alone, especially for segmentation, in the two related tasks of dialog act segmentation and DA classification for speech from the ICSI Meeting Corpus.
Approaches and Applications of Audio Diarization
160 Citations2006D.A. Reynolds, Pedro A. Torres‐Carrasquillo
An overview of current audio diarization approaches is provided and performance and potential applications are discussed, as well as the performance of current systems as measured in the DARPA EARS Rich Transcription Fall 2004 (RT-04F) speaker diarized evaluation.
Edit detection and parsing for transcribed speech
158 Citations2001Eugene Charniak, Mark Johnson
A simple architecture for parsing transcribed speech in which an edited-word detector first removes such words from the sentence string, and then a standard statistical parser trained on transcribedspeech parses the remaining words is presented.
Getting more mileage from web text sources for conversational speech language modeling using class-dependent mixtures
135 Citations2003Ivan Bulyko, Mari Ostendorf +1 more
This paper shows how training data can be supplemented with text from the web filtered to match the style and/or topic of the target recognition task, but also that it is possible to get bigger performance gains from the data by using class-dependent interpolation of N-grams.
Deterministic parsing of syntactic non-fluencies
133 Citations1983Donald Hindle
It is a mystery that people have little difficulty understanding the non-fluent speech that is the essential medium of everyday life and that children can succeed in acquiring the grammar of a language on the basis of evidence provided by a mixed set of apparently grammatical and ungrammatical strings.
IEEE Transactions on Speech and Audio ProcessingAutomatic Recognition of Spontaneous Speech for Access to Multilingual Oral History Archives
133 Citations2004Bill Byrne, David Doermann +10 more
Results are presented from experiments with speech recognition, topic segmentation, topic categorization, and named entity detection using a large collection of recorded oral histories to evaluate the degree to which automatic speech recognition (ASR)-based segmentation and categorization techniques can be adapted to approximate decisions made by human annotators.
Lecture notes in computer scienceProsody in Speech Understanding Systems
117 Citations1997Ralf Kompe
The baseline ASU systems, including prosodic labeling and speech corpora, and prosodic phrase models, suggest that prosodic attributes in ASU are likely to be related to language quality.
IEEE International Conference on Acoustics Speech and Signal ProcessingConnectionist language modeling for large vocabulary continuous speech recognition
116 Citations2002Holger Schwenk, Jean‐Luc Gauvain
The connectionist language model is being evaluated on the DARPA HUB5 conversational telephone speech recognition task and preliminary results show consistent improvements in both perplexity and word error rate.
Edinburgh Research Explorer (University of Edinburgh)The Web as a Baseline: Evaluating the Performance of Unsupervised Web-based Models for a Range of NLP Tasks
96 Citations2004Mirella Lapata, Frank Keller
For the majority of tasks, simple, unsupervised models perform better when n-gram frequencies are obtained from the web rather than from a large corpus, and it is argued that web-based models should be used as a baseline for, rather than an alternative to, standard models.
North American Chapter of the Association for Computational LinguisticsImproved Machine Translation Performance via Parallel Sentence Extraction from Comparable Corpora
93 Citations2004Dragos Stefan Munteanu, Alexander Fraser +1 more
IEEE Transactions on Speech and Audio ProcessingVERBMOBIL: the use of prosody in the linguistic components of a speech understanding system
90 Citations2000Elmar Nöth, Anton Batliner +3 more
It is shown how prosody can be used in speech understanding systems and the VERBMOBIL speech to-speech translation system is demonstrated, which is the first complete system which successfully uses prosodic information in the linguistic analysis.
Dialog Act Tagging Using Graphical Models
77 Citations2006Gang Ji, Jeffrey A. Bilmes
Tests on the ICSI meeting recorder dialog act (MRDA) corpus show that the factored language model implementations are better than the switching n-gram approach and that by using virtual evidence, the label bias problem in conditional models can be avoided.
From Text to Speech Summarization
71 Citations2006Kathleen McKeown, Julia Hirschberg +2 more
It is illustrated how features derived from speech can help determine summary content within two ongoing summarization projects at Columbia University.
Structural Metadata Research in the EARS Program
63 Citations2006Y. Liu, E. Shriberg +8 more
A representative sample of results shows that combining multiple knowledge sources (words, prosody, syntactic information) is helpful, that prosody is more helpful for news speech than for conversational speech, that word errors significantly impact performance, and that discriminative models generally provide benefit over maximum likelihood models.
SRI's 2004 NIST Speaker Recognition Evaluation System
50 Citations2006Sachin Kajarekar, Luciana Ferrer +5 more
A system that uses discriminant features from cepstral coefficients, and systems that use discriminant models from word n-grams and syllable-based NERF n- grams together with a cEPstral baseline system are evaluated.
Archive ouverte UNIGE (University of Geneva)Multi-level Dialogue Act Tags
44 Citations2004Alexander Clark, Andréi Popescu-Belis
MALTUS, a new tagset based on the ICSI-MR and Switchboard tagsets, which satisfies the requirements of multi-layered tagsets for dialogue acts and attempts to compare the merits of integrated versus multi-level classifiers for the detection of dialogue acts.
Combining Text and Audio-Visual Features in Video Indexing
42 Citations2006Shih‐Fu Chang, R. Manmatha +1 more
How imperfect text data obtained by automatic speech recognition (ASR) may be used to help solve challenging problems, such as story segmentation, concept detection, retrieval, and topic clustering are focused on.
Measuring Human Readability of Machine Generated Text: Three Case Studies in Speech Recognition and Machine Translation
36 Citations2006Douglas A. Jones, Edward Gibson +5 more
Three experiments that test the readability of current state-of-the art system output from an automated English speech-to-text (SST) system, applying standard psycholinguistic testing procedures and a modified version of the standard defense language proficiency test for Arabic, learn that subjects are slowed down by about 25% when reading system STT output.
Machine Translation in the Year 2004
35 Citations2006Kai D. Knight, Daniel Marcu
The basic elements of state-of-the-art, statistical MT, including modeling, decoding, evaluation, and data preparation are covered.
Parsing conversational speech using enhanced segmentation
28 Citations2004Jeremy G. Kahn, Mari Ostendorf +1 more
It is demonstrated that a state-of-the-art segmenter, relative to a pause-based segmenting, gives more than 45% of the possible error reduction in parser performance, and that presentation of interruption points to the parser improves performance over using sentence boundaries alone.
Sentence-Internal Prosody Does not Help Parsing the Way Punctuation Does
24 Citations2004Michelle Gregory, Mark Johnson +1 more
The accuracy of a statistical parser on the LDC Switchboard treebank corpus of transcribed sentence-segmented speech is compared using various combinations of punctuation and sentence-internal prosodic information (duration, pausing, and f0 cues).
Computer Speech & LanguageImproving out-of-vocabulary name resolution
19 Citations2004David D. Palmer, Mari Ostendorf
Algorithms for generating targeted name lists for candidate out-of-vocabulary words for applications in language processing, particularly speech recognition, and a simple variation of the approach can be used to generate alternate name spellings, which may be useful for query expansion in information retrieval.
Information Extraction
18 Citations2006Lance Ramshaw, Ralph Weischedel
Information extraction maps a language stream into database records that capture part of its meaning, and a wide variety of learning techniques for these tasks are explored.
Portability Challenges in Developing Interactive Dialogue Systems
16 Citations2006Yuqing Gao, Liang Gu +1 more
Challenges as well as potential solutions are discussed in several critical issues of efficient language modeling, utilization of untranscribed speech data, automatic annotation, and cross-lingual modeling.
Parsing and Its Applications for Conversational Speech
13 Citations2006Matthew Lease, Eugene Charniak +1 more
This paper provides an introduction to recent work in statistical parsing and its applications for conversational speech, with particular emphasis on the relationship between parsing and detecting speech repairs.
User-centered Evaluation for Machine Translation of Spoken Language
11 Citations2006David D. Palmer
A user-centered method for qualitative comparison of machine translation systems based on rankings of system output that can be performed by monolingual users with no training in machine translation evaluation is discussed.
Automatically incorporating unknown words in JUPITER
9 Citations2000Grace Chung
This paper will detail the development of a system that uses an ANGIE-based framework to model both spelling and pronunciation simultaneously, and uses automatically derived novel lexical units in the first stage of the three-stage system.
Real-World Audio Indexing Systems
7 Citations2006Barry K. Logan, David Goddeau +1 more
The need for and description of the key components of real world audio indexing systems are motivating and the various flavors of such systems, the advantages and disadvantages of each, user interfaces, system architectures and evaluation issues are described.