CV
Langchu Huang
Education
- B.E. in Computer Science and Technology2022.09 — 2026.08College of Engineering, Southern University of Science and Technology (SUSTech)GPA: 3.8 / 4.0 (91 / 100)Advisor: Prof. Yang XuSelected Courses: NLP (A+), ML (A), DL (A-), AI (A-)Research Interest: Music Information Retrieval, Natural Language Processing, Symbolic Music
- Non-degree in Exchange Program2025.09 — 2025.12School of Engineering, Hong Kong University of Science and Technology (HKUST)Selected Courses: Elective coursework in music theory and composition
Experience
- Undergraduate Research2024.09 — 2024.12College of Engineering, Southern University of Science and TechnologyInvestigated how paralinguistic signals in dialogue corpora influence the representation learning and downstream task performance of fine-tuned language models.
- Department of Electrical and Computer Engineering, North Carolina State UniversityConducted research in the ARoS Lab under Prof. Edgar Lobaton. Designed a data collection protocol and built a dual-modal cough-speech dataset using both air and bone conduction microphones. Developed a dual-path contrastive learning architecture for robust cough-speech detection.
- Visiting Research Student2025.08 — 2025.11School of Engineering, Hong Kong University of Science and TechnologyAddressed long-context challenges in multitrack symbolic music modeling by developing a multitrack symbolic music tokenizer to design compact discrete representations for symbolic music understanding and generation. The work explored residual vector quantization, interleaved ABC notation, and multimodal alignment for controllable, musically coherent outputs.
- Research Intern2025.11 — 2026.05Lyra Lab, Tencent Music Entertainment
- Developed LyricMidi, a lyric-aware symbolic music foundation model with attribute-wise masked pretraining and hybrid music-language embeddings, improving vocal-centric understanding while maintaining strong performance on symbolic music classification benchmarks.
- Developed symbolic accompaniment generation systems that transform lead sheets into full piano arrangements and guitar tablature.
- Pretrained a unified symbolic music foundation model across performance / score MIDI and ABC notation to support cross-modal downstream tasks such as transcription, PM2S, and conditional generation.
- Developed an end-to-end generative PM2S system for translating performance MIDI into sheet music.
Publications
- Investigating the Representation of Backchannels and Fillers in Fine-tuned Language Models2026ACL 2026 Main ConferenceAuthors: Yu Wang, Leyi Lao, Langchu Huang, Gabriel Skantze, Yang Xu, Hendrik BuschmeierThis work investigates how fine-tuning strategies enhance language models’ ability to represent and generate backchannels and fillers in dialogue. Empirical analyses demonstrate improved clustering quality and more human-like language generation, showcasing the potential for adapting general LMs into pragmatic-aware conversational models.
- LyricMidi: Joint Music–Language Pretraining for Symbolic Music Understanding2025PreprintAuthors: Langchu Huang, Yiliang Jiang, Ying Chen, Lincheng Kong, Weifeng ZhaoThis paper proposes a unified music–language pretraining framework that jointly models symbolic music events and lyrics. The model achieves consistent improvements on vocal-centric symbolic music understanding tasks while preserving general music representation performance.
Awards
- Academic Scholarship, SUSTech2022, 2023, 2024
- Outstanding Graduate, SUSTech2026
- First Prize, 7th National Student Art Exhibition, for Mahler Piano Quartet in A Minor2023
- Second Prize, Chamber Music (Non-Major Division), 3rd GBA College Student Art Festival2024
Certificates
- TOEFL (MyBest)100 (R 25, L 27, S 23, W 25)
- GRE317 (V 152, Q 165)
Portfolio
- Chamber music performance as pianist in Gustav Mahler’s Piano Quartet in A Minor.
- Designed and implemented an extension to integrate school academic services into developers' workflow.
- A dual-path health monitoring system using both air conduction and bone conduction audio signal. Developed during internship at NCSU ECE.
Skills
- Programming : Python, C++, TypeScript, LaTeX, Lua
- Music : Piano, Chamber Music Performance, Music Theory and Notation
Languages
Cantonese (native), Mandarin (native), English (fluent)
Teaching
- Acoustics and Music Teaching for Rural Children2023.07Volunteer Teacher in Lianping Teaching Support Program, providing introductory acoustics and music lessons to rural primary school students
- Volunteer Guide and Interpreter2025.06University of San Francisco Shenzhen Summer Program
- Piano and Music Theory Tutoring2024, 2025Provided individualized private instruction to children and adolescents for 2 years, focusing on fundamental to intermediate performance skills and music literacy
Leadership
- President, MovieSalon Club2023.07 - 2024.07SUSTech
- President, Camerata Coffee Club2023.07 - 2024.07SUSTech