madmom
Python MIR library with neural-network models for beat, downbeat and onset detection.
| Language | Python |
|---|---|
| Category | Audio analysis |
| License | BSD-2-Clause |
| Platforms | macOS Windows Linux |
| Install | pip install madmom |
| First released | ~2015 |
| Maintained | Not actively maintained (last activity 2024) |
| Links | Website · Source |
Strengths
- Strong pretrained RNN/CNN models for beat & downbeat tracking
- Focused on music information retrieval
- Well-cited in academic MIR work
Limitations
- Low maintenance: last release 2018
- Model/data files are non-commercial (CC-BY-NC-SA)
Best for
Research-grade beat, downbeat, and onset detection with pretrained models.
Minimal example
from madmom.features.beats import RNNBeatProcessor
proc = RNNBeatProcessor()
act = proc('audio.wav')