Madmom

Madmom is a specialized Python library designed for audio signal processing with a core focus on Music Information Retrieval (MIR). Architecturally, it integrates advanced signal processing algorithms with deep learning models, particularly Recurrent Neural Networks (RNNs) and Convolutional Neural Networks (CNNs), to perform high-precision tasks such as beat tracking, onset detection, and chord recognition. As of 2026, Madmom remains a foundational pillar in the MIR community, bridging the gap between raw signal analysis and high-level musical semantic understanding. It utilizes Hidden Markov Models (HMMs) for temporal decoding, providing some of the most accurate tempo estimation and beat-tracking capabilities available in the open-source ecosystem. The library is optimized for researchers and developers building music-tech applications, including automated transcription, rhythm analysis, and intelligent audio editing. While newer transformer-based models are emerging, Madmom’s computational efficiency and proven reliability in real-time or batch-processing environments ensure its continued dominance in industrial-grade audio pipelines and academic research environments.