About Me
Hi there! My name is Hao Hao Tan (郑豪好), or you can call me Harry. I am from Kuala Lumpur, Malaysia, and currently residing in Singapore.
I’m currently at BandLab Technologies, working on AI-assisted features to enhance music creation workflow. You can have a look at a list of our projects here. We make a bunch of cool stuffs for music making, such as SongStarter, Voice Changer, Splitter, Smart MIDI tools (Copilot for MIDI), Audio-to-MIDI, etc. On the nerdy side, I also work on ML engineering topics like on-device model serving and optimization – some of the features above run directly on users’ mobile device!
My experience centers on technologies that enable creative applications. While studying at NTU Singapore, I wrote my bachelor’s thesis on AI music generation using deep learning (LSTMs at the time). This work led to a presentation at the ICCC conference and, after graduation, a position as a research assistant in Prof. Dorien Herremans’ lab, where I focused on controllable music generation and audio synthesis. I later joined a music tech startup called Fairphonic, where I developed a deep learning–based music content ID system capable of identifying both original tracks and cover versions to improve royalty attribution. Before joining BandLab, I worked at TikTok on the deep ranking module of its feed recommendation system.
Sometimes I also work on independent research / open-source projects, such as:
- Syntheon, a parameter inference model for synths;
- frechet-audio-distance, a lightweight library for Frechet Audio Distance calculation.
- A PyTorch implementation of differentiable wavetable synthesis.
I am a classically trained pianist, and I listen to all kinds of music (Cantonese classics, IU, Rush Garcia, TheFatRat, Vicetone, Glacier, Avenza, 裁缝铺, Drics…). If you are looking for some cool music, feel free to listen to these playlists that I made!
Please feel free reach me at helloharry66
[at] gmail
or @GoodGood014 on Twitter.