SonicMaster: Towards Controllable All-in-One Music Restoration and Mastering
cs.SD, cs.AI, cs.MM, eess.AS
Submitted: 2025-08-05
Updated: 2026-09-11
Journal ref: Proceedings of ICML, 2026, Seoul, South Korea
Code: https://github.com/AMAAI-Lab/SonicMaster
License: http://creativecommons.org/licenses/by-sa/4.0/
The gist: Music recordings often suffer from audio quality issues such as excessive reverberation, distortion, clipping, tonal imbalances, and a narrowed stereo image, especially when created in
Terminology
Abstract
Music recordings often suffer from audio quality issues such as excessive reverberation, distortion, clipping, tonal imbalances, and a narrowed stereo image, especially when created in non-professional settings without specialized equipment or expertise. These problems are typically corrected using separate specialized tools and manual adjustments. In this paper, we introduce SonicMaster, the first unified generative model for music restoration and mastering that addresses a broad spectrum of audio artifacts with text-based control. SonicMaster is conditioned on natural language instructions to apply targeted enhancements, or can operate in an automatic mode for general restoration. To train this model, we construct the SonicMaster dataset, a large dataset of paired degraded and high-quality tracks by simulating common degradation types with nineteen degradation functions belonging to five enhancements groups: equalization, dynamics, reverb, amplitude, and stereo. Our approach leverages a flow-matching generative training paradigm to learn an audio transformation that maps degraded inputs to their cleaned, mastered versions guided by text prompts. Objective audio quality metrics demonstrate that SonicMaster significantly improves sound quality across all artifact categories. Furthermore, subjective listening tests confirm that listeners prefer SonicMaster's enhanced outputs over other baselines.
Sources
- ImprovNet -- Generating Controllable Musical Improvisations with Iterative Corruption Refinement
- Dual-Path Transformer Network: Direct Context-Aware Modeling for End-to-End Monaural Speech Separation
- Hybrid Spectrogram and Waveform Source Separation
- FLUX that Plays Music
- Classifier-Free Diffusion Guidance
- TangoFlux: Super Fast and Faithful Text to Audio Generation with Flow Matching and Clap-Ranked Preference Optimization
- Distortion Audio Effects: Learning How to Recover the Clean Signal
- Searching For Music Mixing Graphs: A Pruning Approach
- MaskSR: Masked Language Model for Full-band Speech Restoration
- VoiceFixer: Toward General Speech Restoration with Neural Vocoder
- Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow
- Zero-Shot Unsupervised and Text-Based Audio Editing Using DDPM Inversion
- Automatic music mixing with deep learning and out-of-domain data
- Automatic Equalization for Individual Instrument Tracks Using Convolutional Neural Networks
- Diffusion-Based Audio Inpainting
- Exploring trends in audio mixes and masters: Insights from a dataset analysis
- Efficient neural networks for real-time modeling of analog dynamic range compression
- Meta Audiobox Aesthetics: Unified Automatic Quality Assessment for Speech, Music, and Sound
- Music Source Restoration
Related papers
- Few-Shot Open-Set Audio Classification via Transductive Prototype Refinement and Class Logit Enhancement
- Spectral Masking and Interpolation Attack (SMIA): A Black-box Adversarial Attack against Voice Authentication and Anti-Spoofing Systems
- AVMeme Exam: A Multimodal Multilingual Multicultural Benchmark for LLMs' Contextual and Cultural Knowledge and Thinking
- SoundWeaver: Compositional Warm-Starting for Text-to-Audio Diffusion Serving
- WASIL: In-the-Wild Arabic Spoken Interactions with LLMs
- Efficient Audiovisual Speech Processing via MUTUD: Multimodal Training and Unimodal Deployment