A5019598209- CCD and CMOS Imaging Sensors
- Image Processing Techniques and Applications
- Natural Language Processing Techniques
- Analytical Chemistry and Sensors
- Advanced Optical Sensing Technologies
- Advanced Memory and Neural Computing
- Infrared Target Detection Methodologies
- Optical Coherence Tomography Applications
- Speech and Audio Processing
- Speech Recognition and Synthesis
- Digital Holography and Microscopy
- Thin-Film Transistor Technologies
- Gas Sensing Nanomaterials and Sensors
- Music and Audio Processing
- Optical measurement and interference techniques
- Topic Modeling
- Text and Document Classification Technologies
Hanyang University
2024
Kyungpook National University
2018-2019
Score-based generative models have shown the real-like quality of synthesized speech in text-to-speech (TTS) area. However, critical artifact score-based is requirement a high computational cost due to iterative sampling algorithm, and it also makes difficult fine-tune TTS-optimized vocoder. In this study, we propose method joint training TTS model HiFi-GAN using compressed log-mel features, guarantees significant even on non-iterative sampling. As result, proposed overcomes some digital...
This paper presents a CMOS image sensor with the in-pixel aperture technique for single-chip 2-D and 3-D imaging. In conventional sensors, is located at camera lens. However, in proposed sensor, integrated on chip formed metal layer of (CIS) process. A pixel array composed W, R, B PA pixels (W aperture) extracting color depth information. While W becomes blurred increasing distance from focused object, maintains sharpness. Therefore, can be obtained using defocus method. The size pixel,...
In this paper, complementary metal oxide semiconductor (CMOS) image sensors that can extract depth information using the pixel aperture technique is presented. The array of proposed sensor composed blue, red, and white pixels, as well apertures. apertures are formed by pattern in pixels. focused defocused images obtained, simultaneously, sensor, used for calculating information. was designed fabricated 0.11-μm CMOS process, its performance evaluated.
Expressive text-to-speech (TTS) aims to synthesize better human-like speech by incorporating diverse styles or emotions. While most expressive TTS models rely on reference condition the style of generated speech, they often fail generate regular quality. To ensure consistent quality, we propose an conditioned representation extracted from text itself. implement this text-based predictor, design a module residual vector quantization. Furthermore, is enhanced through style-to-text alignment...
Conversational text-to-speech (TTS) aims to synthesize natural voices appropriate a situation by considering the context of past conversations as well current text. However, analyzing and modeling conversation remains challenging. Most conversational TTS use content historical recent without distinguishing between them often generate speech that does not fit situation. Hence, we introduce novel TTS, H4C-TTS, leverages multi-modal realize contextually synthesis. To facilitate modeling, design...
This paper presents the effects of aperture diameter on image blur complementary metal-oxide-semiconductor (CMOS) sensor with pixel apertures for depth extraction. In a conventional camera system, is located at lens. However, in proposed photodiode. The patterns array fabricated CIS are composed blue, red, and white pixels, as well apertures. formed by metal pattern pixels designed using layer process. focused defocused images simultaneously obtained without can be used reference to extract...
In this paper, a wide dynamic range complementary metal-oxide-semiconductor (CMOS) image sensor with the adjustable sensitivity by using cascode field-effect transistor (MOSFET) and inverter is proposed. The characteristics of CMOS were analyzed through experimental results. proposed active pixel consists eight transistors operated under various light intensity conditions. MOSFET as constant current source. generated from varies intensity. has high illumination owing to logarithmic response...
A complementary metal oxide semiconductor (CMOS) binary image sensor is proposed for low-power and low-noise operation. The has the advantages of reduced power consumption fixed pattern noise (FPN). gate/body-tied (GBT) p-channel metal-oxide-semiconductor field-effect transistor (PMOSFET)-type photodetector used as CMOS sensor. GBT PMOSFET-type a floating gate that amplifies photocurrent generated by incident light. Therefore, sensitivity higher than other photodetectors. consists pixel...
Effects of aperture size on the performance CMOS image sensor with pixel for depth extraction are investigated. In general, is related to resolution and sensitivity sensor. As decreases, improved decreases. To optimize size, optical simulation using finite-difference time-domain method was implemented. The performed various sizes from 0.3 μm 1.1 power incidence angle as a function evaluated. Based results, designed fabricated 0.11 process. effects investigated by comparison measurement results.