Neural Audio Codec & Compression

12 repos

Deep learning approaches to audio encoding and compression that balance quality, bitrate, and computational efficiency. These repositories implement neural codec architectures—primarily using transformers and PyTorch—to achieve high-fidelity audio compression at extremely low bitrates (often under 10 kbps). The cluster spans both foundational codec frameworks like EnCodec and domain-specific variants optimized for different sample rates and bandwidth constraints, making it relevant for streaming, communication, and embedded audio applications.

Related clusters