Automatic Music Mixing using a Generative Model of Effect Embeddings
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914150657032192 |
|---|---|
| author | Moliner, Eloi Martínez-Ramírez, Marco A. Koo, Junghyun Liao, Wei-Hsiang Cheuk, Kin Wai Serrà, Joan Välimäki, Vesa Mitsufuji, Yuki |
| author_facet | Moliner, Eloi Martínez-Ramírez, Marco A. Koo, Junghyun Liao, Wei-Hsiang Cheuk, Kin Wai Serrà, Joan Välimäki, Vesa Mitsufuji, Yuki |
| contents | Music mixing involves combining individual tracks into a cohesive mixture, a task characterized by subjectivity where multiple valid solutions exist for the same input. Existing automatic mixing systems treat this task as a deterministic regression problem, thus ignoring this multiplicity of solutions. Here we introduce MEGAMI (Multitrack Embedding Generative Auto MIxing), a generative framework that models the conditional distribution of professional mixes given unprocessed tracks. MEGAMI uses a track-agnostic effects processor conditioned on per-track generated embeddings, handles arbitrary unlabeled tracks through a permutation-equivariant architecture, and enables training on both dry and wet recordings via domain adaptation. Our objective evaluation using distributional metrics shows consistent improvements over existing methods, while listening tests indicate performances approaching human-level quality across diverse musical genres. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_08040 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Automatic Music Mixing using a Generative Model of Effect Embeddings Moliner, Eloi Martínez-Ramírez, Marco A. Koo, Junghyun Liao, Wei-Hsiang Cheuk, Kin Wai Serrà, Joan Välimäki, Vesa Mitsufuji, Yuki Audio and Speech Processing Sound Music mixing involves combining individual tracks into a cohesive mixture, a task characterized by subjectivity where multiple valid solutions exist for the same input. Existing automatic mixing systems treat this task as a deterministic regression problem, thus ignoring this multiplicity of solutions. Here we introduce MEGAMI (Multitrack Embedding Generative Auto MIxing), a generative framework that models the conditional distribution of professional mixes given unprocessed tracks. MEGAMI uses a track-agnostic effects processor conditioned on per-track generated embeddings, handles arbitrary unlabeled tracks through a permutation-equivariant architecture, and enables training on both dry and wet recordings via domain adaptation. Our objective evaluation using distributional metrics shows consistent improvements over existing methods, while listening tests indicate performances approaching human-level quality across diverse musical genres. |
| title | Automatic Music Mixing using a Generative Model of Effect Embeddings |
| topic | Audio and Speech Processing Sound |
| url | https://arxiv.org/abs/2511.08040 |