A Speech Algorithm Engineer at NetEase Cloud Music π΅, passionate about TTS, Voice Conversion, and AI Music Generation. Fudan University M.S. graduate with hands-on experience in speech synthesis, voice cloning, and cutting-edge audio AI research π.
π΅ Currently a Speech Algorithm Engineer at NetEase Cloud Music, where I work on TTS, Voice Conversion, AI Music Generation & Evaluation, and MV video automation. I hold a Master’s degree in Electronic Information from Fudan University, with deep expertise in speech synthesis, voice cloning, and audio generation.
π‘ Previously interned at ByteDance and Bilibili, diving into training-free diffusion acceleration, CV/NLP applications, and ComfyUI plugin development. I love pushing the boundaries of AI in speech, music, and audio β turning research ideas into real-world products.
π Outside of work, I’m a fan of ultimate frisbee and running. Music-wise, I’m into YOASOBI and Jay Chou. Also a big mystery & detective film/TV enthusiast π.
π οΈ Here are a few technologies I've been working with recently:π€ Leading TTS & Voice Conversion R&D β improving naturalness and expressiveness of speech synthesis. πΆ Building AI music generation algorithms, exploring generative models for creative music production. π Designing AI music evaluation systems that align generated music quality with human aesthetics. π¬ Developing Agents for automated song MV generation from end to end.
β‘ Optimized DDPM inference speed through improved distillation strategies. π¬ Deep-dived into Classifier-Free Guidance (CFG), tracking RCFG, Limited Interval and other cutting-edge improvements. π Independently developed a training-free CFG acceleration algorithm β achieving 20% speedup with quality parity. 𧩠Shipped a ComfyUI plugin integrating self-developed CFG acceleration with other train-free methods.
π― Trained models for low-quality & traffic-directing video recognition across the platform. π€ Designed prompt-based title analysis and built a BERT-powered video classification system. πΉ Explored AIGC-driven video production β automated news extraction, image/audio pairing, and video generation.
π‘ Researched diffusion model applications on SAR images at the School of Information Science and Engineering. Published 3 first-author papers and received multiple honors during my studies.
π Studied Cyberspace Security with a focus on Image Captioning algorithm optimization & acceleration. Active in student leadership and competitions.