sp-nitech/diffsptk

A differentiable version of SPTK

59
/ 100
Established

Implements classic speech processing algorithms (STFT, mel-cepstral analysis, LPC, WORLD vocoder components) as differentiable PyTorch layers, enabling end-to-end optimization of signal processing pipelines within neural networks. Built on PyTorch 2.3.1+, it supports both object-oriented modules and functional APIs for flexible integration with audio synthesis and analysis tasks. Covers specialized operations like pitch extraction, spectral envelope analysis (CheapTrick), aperiodicity estimation (D4C), and polyphase quadrature mirror filter banks for subband decomposition.

196 stars. Available on PyPI.

Maintenance 10 / 25
Adoption 10 / 25
Maturity 25 / 25
Community 14 / 25

How are scores calculated?

Stars

196

Forks

20

Language

Python

License

Apache-2.0

Last pushed

Feb 26, 2026

Commits (30d)

0

Dependencies

11

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/sp-nitech/diffsptk"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.