Comment SOUND to get the link What Meta quietly shipped here is prompt-based audio segmentation, not just noise removal. You’re not picking tracks or tweaking sliders. You’re asking the model to understand the sound scene. It works on messy real-world audio, handles overlapping sounds, and doesn’t require clean stems or timestamps. Upload once, describe once, export. No credits, no watermark, no “free for now” limits. If you work with reels, podcasts, voice notes, or found footage, this saves hours you didn’t know you were wasting. #aitools #aicommunity #meta #samaudio #aisound