astorfi/speechpy

Provides mel-frequency cepstral coefficients (MFCCs), mel-filterbank energies, and log-filterbank energies through configurable frame-based feature extraction with adjustable FFT length, filter counts, and frequency bands. The library implements standard speech signal processing pipelines using NumPy, performing frame windowing and Mel-scale spectral analysis to produce feature matrices suitable for machine learning models. Designed as a standalone Python package with no external ML framework dependencies, making it flexible for integration with any downstream speech recognition or audio classification system.

886 stars and 2,224 monthly downloads. No commits in the last 6 months. Available on PyPI.