Spoken Language Generation and Translation 2
HomeFull List of Titles
1: ICSLP'98 Proceedings
Keynote SpeechesText-To-Speech Synthesis 1Spoken Language Models and Dialog 1Prosody and Emotion 1Hidden Markov Model Techniques 1Speaker and Language Recognition 1Multimodal Spoken Language Processing 1Isolated Word RecognitionRobust Speech Processing in Adverse Environments 1Spoken Language Models and Dialog 2Articulatory Modelling 1Talking to Infants, Pets and LoversRobust Speech Processing in Adverse Environments 2Spoken Language Models and Dialog 3Speech Coding 1Articulatory Modelling 2Prosody and Emotion 2Neural Networks, Fuzzy and Evolutionary Methods 1Utterance Verification and Word Spotting 1 / Speaker Adaptation 1Text-To-Speech Synthesis 2Spoken Language Models and Dialog 4Human Speech Perception 1Robust Speech Processing in Adverse Environments 3Speech and Hearing Disorders 1Prosody and Emotion 3Spoken Language Understanding Systems 1Signal Processing and Speech Analysis 1Spoken Language Generation and Translation 1Spoken Language Models and Dialog 5Segmentation, Labelling and Speech Corpora 1Multimodal Spoken Language Processing 2Prosody and Emotion 4Neural Networks, Fuzzy and Evolutionary Methods 2Large Vocabulary Continuous Speech Recognition 1Speaker and Language Recognition 2Signal Processing and Speech Analysis 2Prosody and Emotion 5Robust Speech Processing in Adverse Environments 4Segmentation, Labelling and Speech Corpora 2Speech Technology Applications and Human-Machine Interface 1Large Vocabulary Continuous Speech Recognition 2Text-To-Speech Synthesis 3Language Acquisition 1Acoustic Phonetics 1Speaker Adaptation 2Speech Coding 2Hidden Markov Model Techniques 2Multilingual Perception and Recognition 1Large Vocabulary Continuous Speech Recognition 3Articulatory Modelling 3Language Acquisition 2Speaker and Language Recognition 3Text-To-Speech Synthesis 4Spoken Language Understanding Systems 4Human Speech Perception 2Large Vocabulary Continuous Speech Recognition 4Spoken Language Understanding Systems 2Signal Processing and Speech Analysis 3Human Speech Perception 3Speaker Adaptation 3Spoken Language Understanding Systems 3Multimodal Spoken Language Processing 3Acoustic Phonetics 2Large Vocabulary Continuous Speech Recognition 5Speech Coding 3Language Acquisition 3 / Multilingual Perception and Recognition 2Segmentation, Labelling and Speech Corpora 3Text-To-Speech Synthesis 5Spoken Language Generation and Translation 2Human Speech Perception 4Robust Speech Processing in Adverse Environments 5Text-To-Speech Synthesis 6Speech Technology Applications and Human-Machine Interface 2Prosody and Emotion 6Hidden Markov Model Techniques 3Speech and Hearing Disorders 2 / Speech Processing for the Speech and Hearing Impaired 1Human Speech ProductionSegmentation, Labelling and Speech Corpora 4Speaker and Language Recognition 4Speech Technology Applications and Human-Machine Interface 3Utterance Verification and Word Spotting 2Large Vocabulary Continuous Speech Recognition 6Neural Networks, Fuzzy and Evolutionary Methods 3Speech Processing for the Speech-Impaired and Hearing-Impaired 2Prosody and Emotion 72: SST Student Day
SST Student Day - Poster Session 1SST Student Day - Poster Session 2Author Index
A B C D E F G H I
J K L M N O P Q R
S T U V W X Y Z
Multimedia Files
A Generic Algorithm for Generating Spoken Monologues
Authors:
Esther Klabbers, IPO, Center for Research on User-System Interaction (The Netherlands)
Emiel Krahmer, IPO, Center for Research on User-System Interaction (The Netherlands)
Mariët Theune, IPO, Center for Research on User-System Interaction (The Netherlands)
Page (NA) Paper number 278
Abstract:
The defining property of a Concept-to-Speech system is that it combines language and speech generation. Language generation converts the input-concepts into natural language, which speech generation subsequently transforms into speech. Potentially, this leads to a more `natural sounding' output than can be achieved in a plain Text-to-Speech system, since the correct placement of pitch accents and intonational boundaries ---an important factor contributing to the `naturalness' of the generated speech--- is co-determined by syntactic and discourse information, which is typically available in the language generation module. In this paper, a generic algorithm for the generation of coherent spoken monologues is discussed, called D2S. Language generation is done by a module called LGM which is based on TAG-like syntactic structures with open slots, combined with conditions which determine when the syntactic structure can be used properly. A speech generation module converts the output of the LGM into speech using either phrase-concatenation or diphone-synthesis.
SL980278.PDF (From Author)
SL980278.PDF (Rasterized)
TOP
On the Use of Automatically Generated Discourse-Level Information in a Concept-to-Speech Synthesis System
Authors:
Janet Hitzeman, Centre for Speech Technology Research, University of Edinburgh (U.K.)
Alan W. Black, Centre for Speech Technology Research, University of Edinburgh (U.K.)
Paul Taylor, Centre for Speech Technology Research, University of Edinburgh (U.K.)
Chris Mellish, Department of Artificial Intelligence, University of Edinburgh (U.K.)
Jon Oberlander, Human Communication Research Centre, University of Edinburgh (U.K.)
Page (NA) Paper number 591
Abstract:
This paper describes the latest version of the SOLE concept-to-speech system, which uses linguistic information provided by a natural language generation system to improve the prosody of synthetic speech. We discuss the types of linguistic information that prove most useful and the implications for text-to-speech systems.
SL980591.PDF (From Author)
SL980591.PDF (Rasterized)
TOP
Learning Phrase-Based Head Transduction Models for Translation of Spoken Utterances
Authors:
Hiyan Alshawi, AT&T Labs (USA)
Srinivas Bangalore, AT&T Labs (USA)
Shona Douglas, AT&T Labs (USA)
Page (NA) Paper number 293
Abstract:
We describe a method for learning head-transducer models of translation automatically from examples consisting of transcribed spoken utterances and reference translations of the utterances. The method proceeds by first searching for a hierarchical alignment (specifically a synchronized dependency tree) of each training example. The alignments produced are optimal with respect to a cost function that takes into account co-occurrence statistics and the recursive decomposition of the example into aligned substrings. A probabilistic head-transducer model is then constructed from the alignments. We report results of applying the method to English-to-Spanish translation in the domain of air travel information and English-to-Japanese translation in the domain of telephone operator assistance. We also report on a variation on this model-construction method in which multi-word pairings are used in the computation of the hierarchical alignments and head transducer models.
SL980293.PDF (From Author)
SL980293.PDF (Rasterized)
TOP
Probabilistic Dialogue Act Extraction for Concept Based Multilingual Translation Systems
Authors:
Toshiaki Fukada, ATR-ITL (Japan)
Detlef Koll, CMU-ISL (USA)
Alex Waibel, CMU-ISL (USA)
Kouichi Tanigaki, ATR-ITL (Japan)
Page (NA) Paper number 657
Abstract:
This paper describes a probabilistic method for dialogue act (DA) extraction for concept-based multilingual translation systems. A DA is a unit of a semantic interlingua and it consists of speaker information, speech act, concept and argument. Probabilistic models for the extraction of speech acts or concepts are trained as speech act or concept dependent word n-gram models. The proposed method is evaluated on DA-annotated English and Japanese databases. The experimental results show that the proposed method gives a better performance compared to the conventional grammar-based approach. In addition, the proposed method is much more robust for erroneous inputs obtained as speech recognition outputs.
SL980657.PDF (From Author)
SL980657.PDF (Rasterized)
TOP
Fast Decoding For Statistical Machine Translation
Authors:
Ye-Yi Wang, Carnegie Mellon University (USA)
Alex Waibel, Carnegie Mellon University (USA)
Page (NA) Paper number 826
Abstract:
We investigated an efficient decoding algorithm for statistical machine translation. Compared to the other algorithms, this new algorithm is applicable to different translation models, and it is much faster. Experiments showed that the algorithm achieved an overall performance comparable to the state of the art decoding algorithms.
SL980826.PDF (From Author)
SL980826.PDF (Rasterized)
TOP
A Japanese-to-English Speech Translation System: ATR-MATRIX
Authors:
Toshiyuki Takezawa, ATR Interpreting Telecommunications Research Laboratories (Japan)
Tsuyoshi Morimoto, Fukuoka University (Japan)
Yoshinori Sagisaka, ATR Interpreting Telecommunications Research Laboratories (Japan)
Nick Campbell, ATR Interpreting Telecommunications Research Laboratories (Japan)
Hitoshi Iida, ATR Interpreting Telecommunications Research Laboratories (Japan)
Fumiaki Sugaya, ATR Interpreting Telecommunications Research Laboratories (Japan)
Akio Yokoo, ATR Interpreting Telecommunications Research Laboratories (Japan)
Seiichi Yamamoto, ATR Interpreting Telecommunications Research Laboratories (Japan)
Page (NA) Paper number 957
Abstract:
We have built a new speech translation system called ATR-MATRIX (ATR's Multilingual Automatic Translation System for Information Exchange). This system can recognize natural Japanese utterances such as those used in daily life, translate them into English and output synthesized speech. This system is running on a workstation or a high-end PC and achieves nearly real-time processing. The current implementation of our system deals with a hotel room reservation task/domain. We plan to develop a bidirectional speech translation system, i.e., Japanese-to-English and English-to-Japanese. We also plan to develop multi-language output functions from ATR-MATRIX (Japanese-to-English, German and Korean) for the international joint experiment of C-STAR II (Consortium for Speech Translation Advanced Research).