| domain | samaudiolab.com |
| summary | SAM-Audio is a fully open-source foundation model developed by Meta AI that enables audio separation using text, visual, or temporal prompts. It utilizes a PE-AV encoder to process multi-modal inputs like audio waveforms, text, video frames, and time spans, encoding them through a Prompt Encoder. A Separation Network then generates separated target audio and a residual remainder, achieving high-fidelity output. Key innovations include unified prompting support across text, visual, and temporal modalities, and a foundation built on large-scale multimodal correspondence learning. A SAM-Audio Judge model assesses separation quality. Users can begin with a three-step process involving installation, model loading, and audio separation. |
| title | SAM-Audio Lab |
| description | Comprehensive guide and resource hub for SAM-Audio. SAM-Audio is a foundation model for isolating any sound in audio using text, visual, or temporal prompts from Meta AI. |
| keywords | audio, model, text, prompting, sound, temporal, separation, prompts, video, quality, using, sounds, judge, foundation, time, segmentation, processing |
| upstreams |
|
| downstreams |
|
| nslookup | A 172.67.139.170, A 104.21.8.152 |
| created | 2025-12-27 |
| updated | 2026-01-06 |
| summarized | 2026-01-07 |
|
|