# SAM-Audio

> Segment and isolate specific sounds or voices using multimodal prompts.

SAM-Audio is an audio segmentation model from Meta designed to isolate targeted sounds from audio and video recordings. It supports multimodal prompting, allowing users to isolate audio elements using text descriptions, visual bounding boxes on video frames, or specific time spans.

Website: https://huggingface.co/facebook/sam-audio-large

## What it is for

- Isolate a specific speaker or voice from a noisy video recording
- Separate individual musical instruments from an audio track using text prompts
- Extract environmental sounds by selecting visual objects in video footage

## Features

- Multimodal prompting via text descriptions, visual bounding boxes, and waveform time spans
- Targeted sound and voice separation from mixed audio
- Integration with video context for object-directed audio extraction

## Pricing

Free model weights on Hugging Face under custom research license

## Details

- Type: AI model
- Category: [AI](https://xiry.app/topics/ai.md)
- Open source: no
- Platforms: Code library, Self-hosted
- For: Audio engineers, video creators, and AI researchers
- Alternative to: LALAL.AI, Audacity noise removal
- Tags: audio, meta, stem separation, multimodal, sound isolation

## Why it was saved

- [@evolving.ai, 22 December 2025](https://www.instagram.com/p/DSkPewZDdb6/): Allows creators and sound engineers to separate targeted audio elements and instruments directly from video context.

## Related

- [TemPolor](https://xiry.app/resources/tempolor): AI platform to generate, search, and score royalty-free music for videos (Web app, Freemium)
- [ElevenLabs](https://xiry.app/resources/elevenlabs): AI voice generator, voice cloning, and conversational agent platform (Web app, Freemium)
- [Suno](https://xiry.app/resources/suno): AI platform that generates complete songs with vocals and instrumentation from text. (Web app, Freemium)
- [Adobe Podcast](https://xiry.app/resources/adobe-podcast): Web-based AI audio recording, editing, and speech enhancement tool (Web app, Freemium)
- [PersonaPlex-7B (nvidia/personaplex-7b-v1)](https://xiry.app/resources/personaplex-7b-nvidia-personaplex-7b-v1): Open full-duplex speech-to-speech conversational AI model by NVIDIA (AI model, Open source)
- [Cartesia Sonic](https://xiry.app/resources/cartesia-sonic): Ultra-low-latency streaming text-to-speech and instant voice cloning model. (AI model, Freemium)

Updated 3 October 2026. Page: https://xiry.app/resources/sam-audio

Make your own library: Xiry takes your saved reels, tweets, Reddit bookmarks and Watch Later videos, makes them searchable and pulls the useful stuff out, like this library. It's invite-only for now; join the waitlist at https://xiry.app/#waitlist.