| title: SonicMaster AI Audio Mastering | |
| emoji: 🎧 | |
| colorFrom: purple | |
| colorTo: indigo | |
| sdk: gradio | |
| sdk_version: 6.0.0 | |
| app_file: app.py | |
| pinned: false | |
| hardware: t4-medium | |
| # 🎧 SonicMaster — AI Audio Mastering Studio | |
| Text-guided music restoration & mastering powered by **SonicMaster** (ICML 2026) | |
| ## Features | |
| - **12 Mastering Presets** — Auto Enhance, Dereverb, Bass Boost, Brighten, etc. | |
| - **Custom Text Prompts** — Describe what you want in natural language | |
| - **Advanced Controls** — Adjust inference steps, guidance scale, chunking | |
| - **GPU Accelerated** — Runs on Hugging Face ZeroGPU (T4) | |
| ## Usage | |
| 1. Upload your audio track (WAV, MP3, FLAC) | |
| 2. Select a preset or write a custom prompt | |
| 3. Click **Master Track** and wait for the GPU to process | |
| ## Model | |
| - **SonicMaster** by AMAAI Lab — first unified generative model for music restoration & mastering | |
| - Uses flow-matching with text-based control for targeted audio enhancement | |
| - Handles reverb, clipping, EQ, dynamics, and stereo imbalances | |
| ## Paper | |
| [arXiv:2508.03448](https://arxiv.org/abs/2508.03448) | |