MS student at Shanghai Jiao Tong University, working with the X-LANCE group on speech and audio processing.
Currently exploring how discrete speech representations interact with language models — trying to figure out what makes a good speech tokenizer and how to build better training data for speech LLMs. Most of my time goes into reading papers, running experiments that take too long, and writing data pipelines.
Things I work with daily: PyTorch, HuggingFace Transformers, Whisper, torchaudio, way too many YAML config files
When I'm not in the lab you'll probably find me at a coffee shop or playing basketball badly.