Multi-Stage Music Source Restoration with BandSplit-RoFormer Separation and HiFi++ GAN
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866918369610956800 |
|---|---|
| author | Morocutti, Tobias Karystinaios, Emmanouil Greif, Jonathan Widmer, Gerhard |
| author_facet | Morocutti, Tobias Karystinaios, Emmanouil Greif, Jonathan Widmer, Gerhard |
| contents | Music Source Restoration (MSR) targets recovery of original, unprocessed instrument stems from fully mixed and mastered audio, where production effects and distribution artifacts violate common linear-mixture assumptions. This technical report presents the CP-JKU team's system for the MSR ICASSP Challenge 2025. Our approach decomposes MSR into separation and restoration. First, a single BandSplit-RoFormer separator predicts eight stems plus an auxiliary other stem, and is trained with a three-stage curriculum that progresses from 4-stem warm-start fine-tuning (with LoRA) to 8-stem extension via head expansion. Second, we apply a HiFi++ GAN waveform restorer trained as a generalist and then specialized into eight instrument-specific experts. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_04032 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Multi-Stage Music Source Restoration with BandSplit-RoFormer Separation and HiFi++ GAN Morocutti, Tobias Karystinaios, Emmanouil Greif, Jonathan Widmer, Gerhard Sound Machine Learning Audio and Speech Processing Music Source Restoration (MSR) targets recovery of original, unprocessed instrument stems from fully mixed and mastered audio, where production effects and distribution artifacts violate common linear-mixture assumptions. This technical report presents the CP-JKU team's system for the MSR ICASSP Challenge 2025. Our approach decomposes MSR into separation and restoration. First, a single BandSplit-RoFormer separator predicts eight stems plus an auxiliary other stem, and is trained with a three-stage curriculum that progresses from 4-stem warm-start fine-tuning (with LoRA) to 8-stem extension via head expansion. Second, we apply a HiFi++ GAN waveform restorer trained as a generalist and then specialized into eight instrument-specific experts. |
| title | Multi-Stage Music Source Restoration with BandSplit-RoFormer Separation and HiFi++ GAN |
| topic | Sound Machine Learning Audio and Speech Processing |
| url | https://arxiv.org/abs/2603.04032 |