Category

AI & Research

Neural audio codecs and machine learning research applied to audio and signal processing.

MRHF-Codec (emcodec)

MRHF-Codec (emcodec)

Research

Novel neural audio codec that achieves +40dB improvement over Meta's DAC on high-frequency reconstruction. MS thesis research.

Language: Python 3.9+Framework: PyTorchArchitecture: Dual-path encoder-decoder with FSQ quantizationTraining: GAN-based (Multi-Period + Multi-Scale Discriminators)
Published Dataset

HTMOneShotLoopClassification

5,561 electronic music samples with 55 extracted audio features. Published on Zenodo with CC-BY-4.0 license. 85.4% baseline accuracy.

DOI: 10.5281/zenodo.17872191
View on Zenodo →
Merged PR

TorchFX: Delay Effect

BPM-synced delay with musical time divisions, ping-pong stereo, and strategy pattern architecture. 82 tests, 1,487 lines.

128-star PyTorch library
View on GitHub →