A5100429732- Phonetics and Phonology Research
- Speech Recognition and Synthesis
- Plant-Microbe Interactions and Immunity
- Plant Pathogenic Bacteria Studies
- COVID-19 and Mental Health
- Speech and Audio Processing
- Sexuality, Behavior, and Technology
- Plant Pathogens and Fungal Diseases
- Advanced Steganography and Watermarking Techniques
- Video Surveillance and Tracking Methods
- Mindfulness and Compassion Interventions
- Mental Health Treatment and Access
- Neurobiology of Language and Bilingualism
- Scientific and Engineering Research Topics
- Artificial Intelligence in Healthcare
- Speech and dialogue systems
- Emotion and Mood Recognition
- Model Reduction and Neural Networks
- FinTech, Crowdfunding, Digital Finance
- Hand Gesture Recognition Systems
- Energy Load and Power Forecasting
- Attachment and Relationship Dynamics
- Education and Vocational Training
- Cinema and Media Studies
- Plant and animal studies
Tencent (China)
2021-2025
Jiangxi University of Finance and Economics
2022
Shanghai University of Political Science and Law
2019-2022
Beijing University of Technology
2019-2021
Union Hospital
2020
Huazhong University of Science and Technology
2020
Union Hospital
2020
Peking University Shenzhen Hospital
2019
Chinese Academy of Forestry
2018
University of Southern California
2017
When native Chinese speakers learn English pronunciation, learners of different ages have pronunciation characteristics. Previous studies on the consonant children mainly focus at one stage, and there is a lack comparative ages, research early childhood relatively scarce. We study errors among in kindergarten through 9th Grade (K-9). divided K-9 into three age groups: Kindergarten Children (KC), Primary-school Students (PS), Middle-school students (MS). The results show: (1) overall...
Among bilingual Mandarin-English children, spoken-English learners of different ages have pronunciation characteristics. This paper studies English vowel errors Chinese from kindergarten to 9th Grade (K-9). We divided K-9 into three age groups: Kindergarten Children (KC), Primary-school Students (PS), and Middle-school students (MS). The results show that: (1) overall acquisition error ratio the KC group is higher than that PS group, group. (2) It easier for master phonemes same in Chinese,...
Live video streaming platforms have attracted millions of streamers and daily active users. For profit popularity accumulation, some mix pornography content into live to avoid online supervision. Therefore, accurate recognition porn in has become a challenging task. Porn present multimodal characteristics including visual acoustic content. streamer method is proposed that uses attention-gated deep features. Our contribution includes the following: (1) features, i.e., spatial, motion audio,...
Multi-modal speech emotion recognition (SER) brings performance improvement compared with single-modal systems. However, the ASR errors from text modality may deteriorate SER performance. This paper proposes an method robust to errors. We explore complementary semantic in-formation audio reduce impact of errors, which is done by attention mechanism calculate weighted acoustic representations. information fused representations hypotheses utilizing adaptive weight, determined auxiliary error...
A secondary analysis investigated the influences of verbal stigmatization from family members on psychological well-being Latino homosexual males. model was proposed to explain effects being exposed labeling stigma messages that supported both Stigma Communication Model and Revised Labelling Theory. Results showed exposure messages, especially those with name calling, associated low self-esteem, perceived social support, highly stigmatized beliefs. Such were also related one’s emotion...
With the advent of era big data, Internet technology has been developed in various industries. As a communications equipment supplier, telecom operators are coming to recognize that Things plays crucial role improving enterprise asset management capabilities and reducing costs. Therefore, on basis machine learning, it is very necessary study risk assessment data IOT credit financial operators. This paper first introduces concept assessment, then studies system, discusses application learning...
Ultra-high density 2.5/3D heterogeneous integration has been considered an essential solution for rebooting computation applications like high performance computing, machine learning, and brain-mimicking neuromorphic in both cloud edge modes. Interposers with active auxiliary circuitry such as tunable power distribution network (PDN), phase locked loops, signaling equalizers/ buffers, even units, are emerging much more attractive flexible platforms than passive interposers these...
Abstract As a special modern accounting object, after years of continuous research on derivative financial instruments, the community has made significant substantive breakthroughs. It become consensus major standard-setting bodies that instruments should be included in statement and measured at fair value. With deepening development market economy advent Internet era, China’s also ushered new period development. Therefore, we combine network technology to break through traditional...
English is a stress-timed language, for which sentence stress or prosodic plays an important role. It's then difficult Chinese who are used to the syllable-timed rhythm learn of [1]. In this paper, we investigate how improve evaluation based on as second language (ESL). Particularly, explore some measures quantify rhythmic differences among (L2) learners stress. To relieve dependency labeled data stress, predict automatically utilizing hierarchical network with bidirectional Long Short-Term...
Abstract Background: Global healthcare systems have been under huge pressure since Coronavirus Disease 2019 (COVID-19) was declared a pandemic. It is critical to prevent further spread of COVID-19 and protect health workers from infection. This study aims figure out short-term physiological impact on induced by working in isolation wards for hours with personal protective equipment (PPE), so as provide insights reducing infection risk workers. Methods: Health who had worked at least five...
From a very brief speech, human listeners can estimate the pitch range of speaker and normalize perception. Spectral features which inherently involve both articulatory phonatory characteristics were speculated to play roles in this process, but few reported directly correlate with speaker’s F0 range. To mimic auditory capability validate speculation, preliminary study we proposed an LSTM-based method from 300 ms-long speech input, turned out outperform conventional method. By two more...
In online English teaching, due to the lack of quick understanding relevant requirements and resulting in low learning efficiency, so problem human-computer interaction teaching is a big problem. The hardware system interactive platform, information collection module, student design interface, adopts structure based on multi-hidden neural network, extracts user from data, realizes intelligent learning, gives corresponding feedback. Through analysis performance algorithm, it proved that...