For prospective members
Current projects
AIM is a research group at Purdue that uses machine learning, signal processing, and probabilistic estimation to build music technology that improves practice and performance for (primarily string) musicians.
Active projects
What you could work on
- Automatic Music Transcription: Streamlining audio-to-MIDI transcription for musicians and educators, including isolating sounds in noisy environments.
- Evaluator Sound: Spectrogram analysis and multi-modal transformers that detect extra, missed, and wrong notes and return an annotated score.
- Evaluator Vision: A mobile app that corrects cellist postural mistakes in real time using on-device computer vision (YOLO and Google MediaPipe).
- Robot Cello: Using reinforcement learning and motion-capture data to teach a robot arm to play cello.
Two earlier projects have concluded but remain documented: Companion and Mus2Vid.
Relevant technologies
Tools of the trade
- Deep learning (pre-trained and self-trained models)
- Fourier analysis and digital signal processing
- Computer vision and pattern recognition
- Reinforcement learning and robotics
Prerequisite skills
What we look for
The only strict requirement is fluency in Python and NumPy, or the ability to get fluent within a week or two of joining, since both are used across every project.
We accommodate almost all skill levels. Our teams are geared toward sophomores and up (including graduate students), but motivated freshmen have joined and thrived. More experienced members regularly work alongside and mentor those with less experience. Being able to work and learn independently is critical to doing research on this team. Knowledge of music theory or performance is helpful but not required.
Students with prior experience in machine learning (PyTorch or TensorFlow), probability and statistics, and applied Fourier analysis tend to make the largest contributions. We are especially glad to find specialists in Windows/macOS software engineering, OpenCV and YOLO, and reinforcement learning and robotics.