• Proceedings of the KSPS conference
• /
• 2003.05a
• /
• pp.83-86
• /
• 2003

This paper presents our style-based language model adaptation for Korean conversational speech recognition. Korean conversational speech is observed various characteristics of content and style such as filled pauses, word omission, and contraction as compared with the written text corpora. For style-based language model adaptation, we report two approaches. Our approaches focus on improving the estimation of domain-dependent n-gram models by relevance weighting out-of-domain text data, where style is represented by n-gram based tf*idf similarity. In addition to relevance weighting, we use disfluencies as predictor to the neighboring words. The best result reduces 6.5％ word error rate absolutely and shows that n-gram based relevance weighting reflects style difference greatly and disfluencies are good predictor.

• Proceedings of the KSPS conference
• /
• 2006.11a
• /
• pp.135-138
• /
• 2006

In this paper, we introduce a method for extracting grapheme-to-phoneme conversion rules from the transcription of speech synthesis database and a prosody modeling method using the light version of ToBI for a Korean conversational style TTS. We focused on representing the characteristics of the conversational speech style and the experimental results show that our proposed methods are suitable for developing a Korean conversional style TTS.

• ETRI Journal
• /
• v.28 no.6
• /
• pp.807-810
• /
• 2006

This letter presents a prediction model for sentence-final intonations for Korean conversational-style text-to-speech systems in which we introduce the linguistic feature of 'modality' as a new parameter. Based on their function and meaning, we classify tonal forms in speech data into tone types meaningful for speech synthesis and use the result of this classification to build our prediction model using a tree structured classification algorithm. In order to show that modality is more effective for the prediction model than features such as sentence type or speech act, an experiment is performed on a test set of 970 utterances with a training set of 3,883 utterances. The results show that modality makes a higher contribution to the determination of sentence-final intonation than sentence type or speech act, and that prediction accuracy improves up to 25% when the feature of modality is introduced.

• The Journal of the Acoustical Society of Korea
• /
• v.23 no.2E
• /
• pp.51-55
• /
• 2004

In this paper, we propose a style-specific language model adaptation scheme using n-gram based tf*idf similarity for Korean spontaneous speech recognition. Korean spontaneous speech shows especially different style-specific characteristics such as filled pauses, word omission, and contraction, which are related to function words and depend on preceding or following words. To reflect these style-specific characteristics and overcome insufficient data for training language model, we estimate in-domain dependent n-gram model by relevance weighting of out-of-domain text data according to their n-. gram based tf*idf similarity, in which in-domain language model include disfluency model. Recognition results show that n-gram based tf*idf similarity weighting effectively reflects style difference.

• Journal of Korea Multimedia Society
• /
• v.25 no.8
• /
• pp.1156-1166
• /
• 2022

In this study, we examined voice shopping interaction in which consumers can be involved in the decision-making process. Sixteen kinds of voice shopping interaction were designed with differences in the existence of screen/product involvement/conversation style. Their effects on trust, cognitive load, satisfaction, and continuous intention to use were evaluated through a survey experiment. The main effect of conversation style was significant, and it was found that the more deeply involved users have higher trust. The interaction effect between conversation style and product involvement was also significant. Low involvement product buyers had the most positive user experience from the conversation style that included 'Ask for preference,' while high involvement product buyers had the most positive user experience from the conversation style that included both 'Ask for preference' and 'Question and Answer.' The main effect and interaction effect of the existence of screen was not significant. The results indicate that a positive user experience can be obtained when users are deeply involved in consumer decision-making, especially in purchasing high-involvement products.

• Phonetics and Speech Sciences
• /
• v.13 no.3
• /
• pp.65-70
• /
• 2021

This paper proposes a method for designing a large recording script for open domain English speech synthesis. For read-aloud style text, 12 domains and 294 sub-domains were designed using text contained in five different news media publications. For conversational style text, 4 domains and 36 sub-domains were designed using movie subtitles. The final script consists of 43,013 sentences, 27,085 read-aloud style sentences, and 15,928 conversational style sentences, consisting of 549,683 tokens and 38,356 types. The completed script is analyzed using four criteria: word coverage (type coverage and token coverage), high-frequency vocabulary coverage, phonetic coverage (diphone coverage and triphone coverage), and readability. The type coverage of our script reaches 36.86% despite its low token coverage of 2.97%. The high-frequency vocabulary coverage of the script is 73.82%, and the diphone coverage and triphone coverage of the whole script is 86.70% and 38.92%, respectively. The average readability of whole sentences is 9.03. The results of analysis show that the proposed method is effective in producing a large recording script for English speech synthesis, demonstrating good coverage in terms of unique words, high-frequency vocabulary, phonetic units, and readability.

• Phonetics and Speech Sciences
• /
• v.2 no.1
• /
• pp.3-11
• /
• 2010

The present study confirmed the intelligibility improvement benefit of clear speech by investigating the intelligibility of Korean stops produced in different speaking styles: conversational, citation-form, and clear speech. This finding supports the Hypo- & Hyper-speech theory that speakers adjust vocal effort to accommodate hearers' speech perception difficulty. A progressive intelligibility improvement was found for the three speaking styles investigated: clear speech was more intelligible than citation-form speech citation-form speech was more intelligible than conversational speech and clear speech was also more intelligible than conversational speech. These findings suggest that the manipulations to elicit three distinct speaking styles in a laboratory setting were successful. Korean lenis stops showed the least intelligibility improvement among the three Korean stop types, and this result suggests that lenis stops should be more resistant to intelligibility enhancement efforts in clear speech than aspirated and fortis stops.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.6
• /
• pp.263-271
• /
• 2008

This paper presents a Korean conversational agent system in a mobile environment using natural language processing techniques. The aim of a conversational agent in mobile environment is to provide natural language interface and enable more natural interaction between a human and an agent. Constructing such an agent, it is required to develop various natural language understanding components and effective utterance generation methods. To understand spoken style utterance, we perform morphosyntactic analysis, shallow semantic analysis including modality classification and predicate argument structure analysis, and to generate a system utterance, we perform example based search which considers lexical similarity, syntactic similarity and semantic similarity.

• The KIPS Transactions:PartB
• /
• v.11B no.1
• /
• pp.91-98
• /
• 2004

This paper proposes how to support various media on online conversational activity using XML(extensible Markup Language). The method converts media information into XML based messages and handles alike conventional text based messages. The XML based messages are unified to an XML document, and then a HTML document is generated using the XML and an XSLT documents in a server. A user in each client can play or present media through the hyperlink that is associated media information on the HTML document. The suggested method supports use of various media (text, image, audio, video, documents, etc) and efficient maintenance of font size, color, and style on messages according to extension and modification of XML tags. For application, this paper implemented the system to support media that has client and server architecture on online conversational activity. A user in each client inputs text or media based message using JAVA applet and servlet on the system, and conversational messages on every users' interfaces are automatically updated whenever a user inputs new message. Media on conversational messages are played or presented according to a user's click on hyperlink. Applications for the media presentation are as follows : distance learning, online game, collaboration, etc.

• Journal of Smart Tourism
• /
• v.3 no.1
• /
• pp.41-47
• /
• 2023

Tourism and hospitality have encountered significant changes in recent years as a result of the rapid development of information technology (IT). Customers now expect more expedient services and customized travel experiences, which has intensified competition among service providers. To meet these demands, businesses have adopted sophisticated IT applications such as ChatGPT, which enables real-time interaction with consumers and provides recommendations based on their preferences. This paper focuses on the AI support-prompt middleware system, which functions as a mediator between generative AI and human users, and discusses two operational rules associated with it. The first rule is the Information Processing Rule, which requires the middleware system to determine appropriate responses based on the context of the conversation using techniques for natural language processing. The second rule is the Information Presentation Rule, which requires the middleware system to choose an appropriate language style and conversational attitude based on the gravity of the topic or the conversational context. These rules are essential for guaranteeing that the middleware system can fathom user intent and respond appropriately in various conversational contexts. This study contributes to the planning and analysis of service design by deriving design rules for middleware systems to incorporate artificial intelligence into tourism services. By comprehending the operation of AI support-prompt middleware systems, service providers can design more effective and efficient AI-driven tourism services, thereby improving the customer experience and obtaining a market advantage.