audioFlux

Core implemented in C with FFT hardware acceleration, enabling efficient processing for large-scale data and real-time mobile audio streams, as highlighted in the benchmark results.

Supports dozens of time-frequency analysis methods including BFT, NSGT, CWT, and reassignment techniques like synchrosqueezing, providing sharp representations for diverse audio tasks.

Offers spectral features, cepstral coefficients (e.g., MFCC), deconvolution, and chroma features for various spectrum types, allowing multi-dimensional combinations tailored to deep learning models.

Runs on Linux, macOS, Windows, iOS, and Android, ensuring broad compatibility for research and production environments across devices.