Filters








36 Hits in 1.4 sec

Automatic Choreography Generation with Convolutional Encoder-decoder Network

Juheon Lee, Seohyun Kim, Kyogu Lee
2019 Zenodo  
Attribution: Juheon Lee, Seohyun Kim, Kyogu Lee.  ...  Lee, Seohyun Kim, Kyogu Lee.  ... 
doi:10.5281/zenodo.3527958 fatcat:jrof3ir4pvg4noip2tfqg46z6i

Listen to Dance: Music-driven choreography generation using Autoregressive Encoder-Decoder Network [article]

Juheon Lee, Seohyun Kim, Kyogu Lee
2018 arXiv   pre-print
Lee et al.'s and Ofil et al.'  ...  Lee et al. proposed a choreography generation algorithm that retrieves the motions corresponding to the most similar pieces of music in the predefined motion-music-paired database for given new music  ... 
arXiv:1811.00818v1 fatcat:g2mrgfx64zc3lpph245n7bqc4i

Expressive Singing Synthesis Using Local Style Token and Dual-path Pitch Encoder [article]

Juheon Lee, Hyeong-Seok Choi, Kyogu Lee
2022 arXiv   pre-print
This paper proposes a controllable singing voice synthesis system capable of generating expressive singing voice with two novel methodologies. First, a local style token module, which predicts frame-level style tokens from an input pitch and text sequence, is proposed to allow the singing voice system to control musical expression often unspecified in sheet music (e.g., breathing and intensity). Second, we propose a dual-path pitch encoder with a choice of two different pitch inputs: MIDI pitch
more » ... sequence or f0 contour. Because the initial generation of a singing voice is usually executed by taking a MIDI pitch sequence, one can later extract an f0 contour from the generated singing voice and modify the f0 contour to a finer level as desired. Through quantitative and qualitative evaluations, we confirmed that the proposed model could control various musical expressions while not sacrificing the sound quality of the singing voice synthesis system.
arXiv:2204.03249v1 fatcat:qb32fmpjircpxoprescktzp3sq

Disentangling Timbre and Singing Style with Multi-singer Singing Synthesis System [article]

Juheon Lee, Hyeong-Seok Choi, Junghyun Koo, Kyogu Lee
2019 arXiv   pre-print
In this study, we define the identity of the singer with two independent concepts - timbre and singing style - and propose a multi-singer singing synthesis system that can model them separately. To this end, we extend our single-singer model into a multi-singer model in the following ways: first, we design a singer identity encoder that can adequately reflect the identity of a singer. Second, we use encoded singer identity to condition the two independent decoders that model timbre and singing
more » ... tyle, respectively. Through a user study with the listening tests, we experimentally verify that the proposed framework is capable of generating a natural singing voice of high quality while independently controlling the timbre and singing style. Also, by using the method of changing singing styles while fixing the timbre, we suggest that our proposed network can produce a more expressive singing voice.
arXiv:1910.13069v1 fatcat:6uujv4fg35d6lbligrmirzeqcy

Adversarially Trained End-to-end Korean Singing Voice Synthesis System [article]

Juheon Lee, Hyeong-Seok Choi, Chang-Bin Jeon, Junghyun Koo, Kyogu Lee
2019 arXiv   pre-print
In this paper, we propose an end-to-end Korean singing voice synthesis system from lyrics and a symbolic melody using the following three novel approaches: 1) phonetic enhancement masking, 2) local conditioning of text and pitch to the super-resolution network, and 3) conditional adversarial training. The proposed system consists of two main modules; a mel-synthesis network that generates a mel-spectrogram from the given input information, and a super-resolution network that upsamples the
more » ... ted mel-spectrogram into a linear-spectrogram. In the mel-synthesis network, phonetic enhancement masking is applied to generate implicit formant masks solely from the input text, which enables a more accurate phonetic control of singing voice. In addition, we show that two other proposed methods -- local conditioning of text and pitch, and conditional adversarial training -- are crucial for a realistic generation of the human singing voice in the super-resolution process. Finally, both quantitative and qualitative evaluations are conducted, confirming the validity of all proposed methods.
arXiv:1908.01919v1 fatcat:322co346tffcvohc6dcajoulj4

Audio Cover Song Identification using Convolutional Neural Network [article]

Sungkyun Chang, Juheon Lee, Sang Keun Choe, Kyogu Lee
2020 arXiv   pre-print
In this paper, we propose a new approach to cover song identification using a CNN (convolutional neural network). Most previous studies extract the feature vectors that characterize the cover song relation from a pair of songs and used it to compute the (dis)similarity between the two songs. Based on the observation that there is a meaningful pattern between cover songs and that this can be learned, we have reformulated the cover song identification problem in a machine learning framework. To
more » ... this, we first build the CNN using as an input a cross-similarity matrix generated from a pair of songs. We then construct the data set composed of cover song pairs and non-cover song pairs, which are used as positive and negative training samples, respectively. The trained CNN outputs the probability of being in the cover song relation given a cross-similarity matrix generated from any two pieces of music and identifies the cover song by ranking on the probability. Experimental results show that the proposed algorithm achieves performance better than or comparable to the state-of-the-art.
arXiv:1712.00166v2 fatcat:rdga2rsglnandptqbbexjl77za

Neural Analysis and Synthesis: Reconstructing Speech from Self-Supervised Representations [article]

Hyeong-Seok Choi, Juheon Lee, Wansoo Kim, Jie Hwan Lee, Hoon Heo, Kyogu Lee
2021 arXiv   pre-print
Inspired by the success of style conversion in computer vision, Chou and Lee [10] also focused on restricting the information flow using instance normalization [44] and adaptive instance normalization  ... 
arXiv:2110.14513v2 fatcat:fvdtpqom4nhwlknlpyqfuionoa

Highly Soluble Fluorinated Polyimides Synthesized with Hydrothermal Process towards Sustainable Green Technology

Juheon Lee, Seungho Baek, Jinsu Kim, Sangrae Lee, Jinyoung Kim, Haksoo Han
2021 Polymers  
Polyimides, a widely used engineering plastic, require use of large amounts of toxic and hazardous organic solvents which threaten our daily lives, calling for new and easy synthetic methods for sustainable environmentally friendly development. In this paper, highly soluble fluorinated polyimides based on 4,4′-(hexafluoroisopropylidene) diphthalic anhydride were synthesized via hydrothermal process without using any toxic organic solvents and the advantages of the newly demonstrated synthetic
more » ... thods are shown by comparative analysis performed with the two conventional synthetic methods using organic solvent: thermal and chemical imidization. Lower temperature is required (~200 °C) compared to thermal imidization and functional groups for high fusibility formed more easily compared to chemical imidization. According to the comparative analysis, hydrothermally synthesized PIs showed excellent solubility and maintained high thermal stability (>500 °C) and glass transition temperature (>300 °C) compared to conventional PI. The hydrothermally synthesized polyimide is much more convenient to store and manage than other form of polyimide which is much more stable when it is exposed to humidity as it is a powder form. The hydrothermal synthetic method is verified to be a "Green" and facile method for sustainable PI synthesis.
doi:10.3390/polym13213824 pmid:34771380 pmcid:PMC8587447 fatcat:nt6bzecupzdzncyjgufasot264

Content-based feature exploration for transparent music recommendation using self-attentive genre classification [article]

Seungjin Lee, Juheon Lee, Kyogu lee
2018 arXiv   pre-print
, Juheon Lee, and Kyogu Lee  ...  www.kaggle.com/gyani95/380000-lyrics-from-metrolyrics/home † http://www.ifs.tuwien.ac.at/mir/msd/TopMAGD.html arXiv:1808.10600v2 [cs.IR] 3 Sep 2018 LBRS@RecSys '18, Vancouver, Canada, October, 2018 Seungjin Lee  ... 
arXiv:1808.10600v2 fatcat:ya7fqg4iwvbs7h673hgjm6aihm

Adversarially Trained End-to-End Korean Singing Voice Synthesis System

Juheon Lee, Hyeong-Seok Choi, Chang-Bin Jeon, Junghyun Koo, Kyogu Lee
2019 Interspeech 2019  
In this paper, we propose an end-to-end Korean singing voice synthesis system from lyrics and a symbolic melody using the following three novel approaches: 1) phonetic enhancement masking, 2) local conditioning of text and pitch to the superresolution network, and 3) conditional adversarial training. The proposed system consists of two main modules; a mel-synthesis network that generates a mel-spectrogram from the given input information, and a super-resolution network that upsamples the
more » ... ed mel-spectrogram into a linear-spectrogram. In the mel-synthesis network, phonetic enhancement masking is applied to generate implicit formant masks solely from the input text, which enables a more accurate phonetic control of singing voice. In addition, we show that two other proposed methods -local conditioning of text and pitch, and conditional adversarial training -are crucial for a realistic generation of the human singing voice in the super-resolution process. Finally, both quantitative and qualitative evaluations are conducted, confirming the validity of all proposed methods.
doi:10.21437/interspeech.2019-1722 dblp:conf/interspeech/LeeCJKL19 fatcat:uz3ekgg24zeerczcewvjmfbx5a

Spec2Spec: Towards the general framework of music processing using generative adversarial networks

Hyeong-Seok Choi, Juheon Lee, Kyogu Lee
2020 Acoustical Science and Technology  
The advent of deep learning has led to a great progress in solving many problems that had been considered challenging. Several recent studies have shown promising results in directly changing the styles between two different domains that share the same latent content, for example, from paintings to photographs and from simulated roads to real roads. One of the key ideas that lie in this series of domain translation approaches is the concept of generative adversarial networks (GANs). Motivated
more » ... this concept of changing a certain style of data into another style using GANs, we apply this technique to two challenging and yet very important applications in the music signal processing field: music source separation and automatic music transcription. Both tasks can be interpreted as a style transition between two different spectrogram domains that share the same content; i.e., from a mixture spectrogram to a specific source spectrogram in the case of source separation, and from an audio spectrogram to a piano roll representation in the case of music transcription. Through experiments using real-world audio, we demonstrate that one general deep learning framework, namely "spectrogram to spectrogram" or "Spec2Spec," can successfully be applied to tackle these problems.
doi:10.1250/ast.41.160 fatcat:fuudsdiqljgxxdgad5baya5vwq

Solventless UV Curable Material for Low Cost System
저에너지 UV 경화형 무용제 소재 개발

KWANGIN KIM, JUHEON LEE, HYUNJU LEE, HAKSOO HAN
2017 Transactions of the Korean hydrogen and new energy society  
>> In this study, Poly-urethane acrylate (PUA) was synthesized by the reaction between Polycaprolactonetriol (PCLT) and Isophorone dissocyanate (IPDI) and hybridized with inorganic materials. Tetraethylortho silicate (TEOS) and nano clay (Closite 20A) were used as inorganic particles. For the hybridization of TEOS with PUA, sol-gel method is used, in which TEOS is made into spherical particle in the firsthand. In the case of Nano clay, hybridization is carried out through the dispersion as Nano
more » ... clay has a layered structure. The solution of PUA hybrid was made into a film after UV curing and its thermo and electrical properties were measured. The experimental analysis and result demonstrate that the PUA hybrid shows an improved thermal properties and lower dielectric constant than that of the non-hybrid PUA. The trend of improved properties was different depending on structure of inorganic materials.
doi:10.7316/khnes.2017.28.1.77 fatcat:drowu4hw5vck5apvuijsfc3obi

Nanodiamond as a Cytokine Sponge in Infectious Diseases

Wonbeak Yoo, Wonhwa Lee, Hong Nam Kim, Jiyoung Jeong, Hee Ho Park, June Hong Ahn, Dana Jung, Juheon Lee, Ji-su Kim, Seung Whan Lee, Wan-Seob Cho, Seokho Kim
2022 Frontiers in Bioengineering and Biotechnology  
Current strategies for CRS treatment include 1) filtration of blood (Frimmel et al., 2014; Greil et al., 2017; Xiao et al., 2019) ; 2) corticosteroids (Lee et al., 2014; Lee et al., 2019) ; and 3) antagonists  ...  Clinical data were collected for all patients and reported in our previous study (Lee et al., 2020b) .  ... 
doi:10.3389/fbioe.2022.862495 pmid:35445003 pmcid:PMC9014093 fatcat:s7kmupn3lvhmneimccmvxbyv6y

Neurovascular coupling under chronic stress is modified by altered GABAergic interneuron activity

Kayoung Han, Jiwoong Min, Myunghee Lee, Bok-Man Kang, Taeyoung Park, Junghyun Hahn, Jaeseung Yei, Juheon Lee, Junsung Woo, C. Justin Lee, Seong-Gi Kim, Minah Suh
2019 Journal of Neuroscience  
et al., 2015; Lee et al., 2018).  ...  Each GABA receptor type has been reported to affect neurovascular coupling (Fergus and Lee, 1997; Jessen et al., 2015) .  ... 
doi:10.1523/jneurosci.1357-19.2019 pmid:31672788 pmcid:PMC6978951 fatcat:vsdin22nrveaxk7lwmmpxdka7m

The Categorization Algorithms Using Edge Detection for Plant Recognition System
외곽선 검출을 이용한 식물자동인식 분류 알고리즘

Juheon Lee, Simong Park, Bomi Kim, Jongmin Lee, Minjung Kim, Aekyung Lee
2014 Journal of Korea Society for Plants People and Environment  
As smart phones are becoming new means of gathering and sharing information, the demands for the smart phone 'apps' providing plant information and data are also increasing. This study is to provide the algorithms to identify plants using their leaves and is also to feature the digitized leaf data to be stored in the database to be used for plant identification applications. Outline 'edge' data of the leaf are extracted by using SP Sweeping algorithm, and the plant species are categorized and
more » ... entified by applying six algorithms: H/W ratio, top-tip ratio, bottom-tip ratio, gradient of edge points, distance between edge point and the center, and second degree polynomial approximation algorithms. Thirty two species have been selected and used to validate the algorithms. H/W ratio and top-tip ratio have been found to be very effective algorithms to distinguish many species. Moreover, gradient and distance methods have shown the recognizable differences between species, and Bhattacharyya distances have been found to be a great data to be stored as a part of analysis data for run-time comparison by application. Additionally, this study also has confirmed the feasibility of additional identification categories using the features of petiole and tip as well as characteristics of the leaf margin.
doi:10.11628/ksppe.2014.17.5.411 fatcat:nasjo6wolncfxnlabwlmzppnvq
« Previous Showing results 1 — 15 out of 36 results