SubRoformer hits SOTA music source separation with DSP subband filters
affige_yang · x · 2026-11-22
SubRoformer (arXiv:2611.08130) by @archimickey replaces Fourier-based frontends with DSP-based subband filters, achieving SOTA music source separation across vocals, bass, drums and other stems.
- Official GitHub repo open-sourced under MIT license, with training code and a real-time separation demo app
- Base model (1.4M updates, EMA) reaches 11.59 mean SDR (12.69 vocals) on the MUSDB18-HQ test set
- Trained on 100 MUSDB18-HQ songs plus 500 internal songs
More from Research
- Lean community certifies Riemann proof; honeypot fake keys expose breaches at three AI providers — LeviTurk · 2026-11-22
- Riemann hypothesis 'proven by Turkish mathematician using AI' claim mocked as the 'Organic Turk' — CatAstro_Piyush · 2026-11-22
- Andriy Burkov releases The Hundred-Page Language Models Book, hands-on with PyTorch — burkov · 2026-11-22
- Generative AI meets cardiology: Dagstuhl seminar bridges ML and clinical practice — StephanMandt · 2026-11-22
- Emily Riehl releases Lean game teaching dependent type theory at intro-to-proofs level — abhi9u · 2026-11-22
- Curated reading list: evaluating mobile-use agents on Android, from AppAgent to AndroidReality — East-Muffin-6472 · 2026-11-22