Auto Masher is a research project that aims to create pop song mashups with AI-assisted music information retrieval and analysis. In this project, we have compiled a dataset of pop songs and their corresponding chords and beats. The user will submit a song from YouTube, and the pipeline will automatically find the best song to mashup with the user's song.
Our technical paper "Retrieval-based automatic mashup generation with deep learning-guided features" has been accepted by the 25th International Congress on Acoustics/188th Meeting of the Acoustical Society of America (ICA2025 New Orleans)
Our paper DOI: https://doi.org/10.1121/2.0002071 has been awarded Best Student Paper by Proceedings of Meetings on Acoustics (POMA)
Auto Masher has been awarded the second-runner up in Best Final Year Project Award in the Department of Computer Science, HKUST in the year 2023-2024.
- Python 3.12 (Ideal version, should theoretically work with Python 3.10 and 3.11, beyond 3.13 the aifc libraries are removed so librosa kinda doesn't work)
- A decent GPU with some VRAM (>= 4GB) if possible
- C++14 compatible compiler (for
madmomlibrary) ffmpeginstalled on your system and added to your PATH
- Clone the repository:
git clone ...
cd AutoMasher- Install the required Python packages:
pip install -r requirements.txt-
Install
mediainfofor metadata extraction: refer to https://mediaarea.net/en/MediaInfo for installation instructions. The binarymediainfoshould be accessible in your system's PATH. -
python main.py