• Proceedings of the Korea Information Processing Society Conference
• /
• 2020.05a
• /
• pp.530-531
• /
• 2020

최근 인공지능을 활용하여 예술 작품에 몰입할 수 있도록 무대 효과를 디자인하는 연구가 진행되고 있다. 무대 효과 중에서 무대 배경은 공연의 분위기를 형성한다. 춤의 장르별로 무대 배경에 사용되는 이미지를 생성하기 위해 소셜 미디어 기반 무대 배경 생성 시스템이 있다. 하지만 같은 장르 춤은 동일한 무대 배경 이미지가 제공되는 문제가 있다. 같은 장르의 춤이지만 노래의 분위기를 반영하여 차별된 무대 배경 이미지를 제공하는 것이 필요하다. 본 논문은 노래 가사의 감정을 활용하여 Generative Adversarial Network(GAN)을 통해 각 노래의 분위기를 고려한 무대 배경 이미지를 생성하는 방법을 제안한다. GAN은 노래에 포함된 단락별 감정 단어를 추출하여 스타일을 생성하도록 학습된다. 학습된 GAN은 노래 가사에 포함된 감정 단어를 활용하여 곡의 분위기를 반영한 무대 배경 이미지를 생성한다. 노래 가사를 고려하여 무대 배경 이미지를 생성함으로써 곡의 분위기가 고려된 무대 배경 이미지 생성이 가능하다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2012.06b
• /
• pp.273-275
• /
• 2012

노래(Song)와 같이 가사를 포함한 음악은 같은 스타일의 멜로디라도 청자에 따라 느끼는 감정이 다르다. 따라서 전통적인 음악 분류에서 사용하는 템포, 박자, 음정, 음표, 리듬과 같은 자질을 이용하여 감정을 분류할 수 없다. 본 연구에서는 가사로부터 감정 자질을 추출하고, 이를 학습 자질로 이용하여 노래 가사의 감정을 분류한다. 감정 자질의 추출 정확도를 높이고자, 한국어의 언어적 특징을 반영한 규칙을 구축한다. 추출된 감정 자질과 structured SVM을 이용하여 노래 가사의 감정을 분류한 결과, Naive Bayes나 SVM과 같은 전통적인 학습 기법보다 높은 성능(accuracy = 68.9%)을 보였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2011.04a
• /
• pp.340-343
• /
• 2011

음악 감정 분류에 관한 기존의 연구들은 템포, 박자, 음정, 음표, 리듬 등과 같은 음악의 멜로디와 관련된 자질을 이용하여 음악 감정을 분류하였다. 그러나 노래(Song)와 같이 가사를 포함한 음악은 같은 스타일의 멜로디라도 가사의 내용에 따라 음악에 대하여 청자가 느끼는 감정이 크게 다르다. 본 논문에서는 감정 온톨로지를 활용하여 노래 가사를 감정에 따라 분류하는 방법에 대하여 제안한다. 기구축 된 감정 온톨로지를 바탕으로 네 가지 통사적 규칙을 적용하여 노래 가사로부터 감정 자질을 추출한다. 추출된 감정 자질을 이용하여 Naive Bayes, HMM, SVM과 같은 기계학습 기법을 이용하여 8개 감정 그룹에 대해 58.8%의 정확도를 보였다.

• Journal of Music and Human Behavior
• /
• v.7 no.2
• /
• pp.23-45
• /
• 2010

The purpose of this study is to determine the effects of lyric analysis using the mind-map technique on parenting stress of mothers with disabled children. Six mothers performed psychological tests to confirm the level of parenting stress. Participants attended a total of fourteen lyric-analysis using mind-map programs and the Questionnaire on Resource and Stress (QRS) was administered before the first session and after the last session. Results showed a significant difference (p<.05) in perceived parenting stress. Also positive changes were observed from video analysis of each session. Overall, this study suggests that the lyric analysis based on the application of mind-map may have the positive influence on the parenting stress of mothers raising disabled children.

• Journal of Music and Human Behavior
• /
• v.12 no.1
• /
• pp.43-64
• /
• 2015

This study investigated the effects of song discussion on depression and rehabilitation motivation in stroke patients. Older adults with chronic stroke participated in this study: nine for the experimental group and eight for the control group. The experimental group was divided into three subgroups and participated in 12 sessions over 6 weeks. Target lyrics were selected by the investigator among popular songs from the participants' young adulthood. The song-based discussion was facilitated to address issues targeted at supportive, insight-focused, or reconstructive stage. The control group was provided with delayed intervention. At pre and posttest, the short form of Geriatric Depression Scale and the Rehabilitation Motivation Scale were measured. The experimental group showed significantly decreased depression and significantly increased rehabilitation motivation (p < .01), while the control group showed no significant changes. Positive changes were also observed in all subcategories of rehabilitation motivation in the experimental group, particularly in significantly increased task-oriented motivation and decreased amotivation. This study suggests that song discussion will be effectively applied in rehabilitative settings to address psychological issues of older adults with stroke.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.26 no.4
• /
• pp.301-307
• /
• 2016

This paper proposes a lyric search method to consider the characteristics of the user query. According to the fact that queries for the lyric search are derived from highlight parts of the music, this paper uses the hierarchical agglomerative clustering to find the highlight and proposes a Gaussian weighting to consider the neighbor of the highlight as well as highlight. By setting the mean of a Gaussian weighting at the highlight, this weighting function has higher weights near the highlight and the lower weights far from the highlight. Then, this paper constructs a index of lyrics with the gaussian weighting. According to the experimental results on a data set obtained from 5 real users, the proposed method is proved to be effective.

• Annual Conference on Human and Language Technology
• /
• 2010.10a
• /
• pp.42-45
• /
• 2010

본 연구에서는 한국의 대중가요의 가사 정보를 형태소 단위로 분석하고 이 정보를 기반으로 노래의 감정을 분류하여 추천하는 시스템을 제안한다. 이 시스템을 구축하기 위해서 수집된 노래의 가사는 형태소를 분석하여 각 형태소를 자질로 결정하고, 사용되는 분류기는 ME 모델을 이용해서 학습된다. 이 학습된 분류기는 자질의 수에 따라 그 성능이 분석되고, 분류기를 사용한 추천 시스템은 랜덤하게 생성된 데이터 집합에 대해서 얼마나 정확하게 노래를 추천하는 지를 분석한다.

• Proceedings of the Korea Contents Association Conference
• /
• 2008.05a
• /
• pp.43-48
• /
• 2008

And this study intends to discuss especially the one using emotional filter among various information processing methods. The existing music recommendation service on the web has a weak point that it makes the user feel bored by recommending songs only with similar feeling of the same genre, because music is classified by tune, melody, atmosphere and genre before recommendation. The service using emotion filter, suggested in this study, recommends the song and lyrics appropriate to the current emotional state of the user by abstracting emotional words that could reflect the sensitivity of human and then search the words within lyrics to match in order to overcome the weak point of the existing service. This study starts where the current emotional status for the user is being input. As for the range to choose, there are the seven representatives of emotion which are, love, separation, joy, sorrow-gloom, happiness-lonesome, and anger. As the service receives input of user's emotion, it matches the emotional words appropriate for the emotion input with the lyrics, and ranks the lyrics in the order of priority, so that it recommends the song and it lyrics to the user.

• Journal of Music and Human Behavior
• /
• v.15 no.2
• /
• pp.41-67
• /
• 2018

The purpose of this case study was to examine how a supportive songwriting program could facilitate resilience for adolescents with school maladjustment. Participants included four middle school students with adaptive difficulties. The participants received eight 50-minute sessions of a supportive songwriting program. The program consisted of singing and discussing selected songs, followed by the participants creating their own lyrics about their current adaptation issues. The Resilience Scale for Adolescents was completed by each participant before and after the intervention, and the participants' lyrics were analyzed for how the factors of resilience were reflected in their lyrics. The results showed that all participants' scores on the resilience scale increased. It was also found that the factor on the resilience scale that increased the most for each participant was related to the issues expressed in their lyrics. The results suggest that the process of writing songs can be effective in eliciting adolescents' school related issues and accessing their positive resources, which can lead to behavioral and psychological improvements.

• KIISE Transactions on Computing Practices
• /
• v.22 no.10
• /
• pp.479-487
• /
• 2016

People can listen to almost every type of music by music streaming services without possessing music. Ironically it is difficult to choose what to listen to. A music recommendation system helps people in making a choice. However, existing recommendation systems have high computation complexity and do not consider context information. Emotion is one of the most important context information of music. Lyrics can be easily computed with various language processing techniques and can even be used to extract emotion of music from itself. We suggest a music-level similarity evaluation method using emotion and structure. Our result shows that it is important to consider semantic information when we evaluate similarity of music.