SAM Audio is an online AI audio separation tool that uses Meta's open-source SAM Audio model to isolate vocals, instruments, and sounds via text, visual, or time-based prompts.
What is SAM Audio?
SAM Audio is a browser-based platform that brings Meta's open-source SAM Audio AI model to users for audio separation. It accepts audio files in formats like MP3, WAV, and FLAC and allows users to describe what they want to isolate using natural language, visual cues from video, or specific time ranges. The output is high-quality separated audio tracks ready for download, all processed in the cloud without requiring any installations.
Key Features
- Multi-modal prompting — Support for text prompts (e.g., "extract the drums"), visual cues from uploaded videos, and time-span selection to target specific moments in the audio.
- Universal audio separation — Isolate vocals, instruments, speech, or environmental sounds from any audio source with state-of-the-art accuracy.
- Music stem separation — Separate vocals, drums, bass, and other instruments from songs with studio-quality clarity.
- Speech enhancement — Extract clean dialogue from noisy recordings, ideal for podcast and video production.
- Background noise removal — Identify and remove unwanted background sounds while preserving the desired audio.
- Browser-based processing — Upload and separate audio directly in the browser; no downloads or installations required.
- Fast cloud processing — Optimized infrastructure delivers separated tracks in seconds, not minutes.
- Multiple output formats — Downloaded tracks are high-quality and ready for professional use.
Who is it for?
- Music producers — Extract individual instruments from reference tracks for remixes, sampling, or arrangement study.
- Podcast editors — Clean up interview recordings by removing background noise or isolating speech.
- Video creators — Separate dialogue from background audio to improve clarity or add effects in video projects.
- Sound designers — Isolate specific sounds from complex audio recordings for use in soundtracks or Foley.
What can you do with SAM Audio?
- Remix and sampling: Extract acapellas and instrumentals from any song to create new mixes or mashups.
- Audio cleanup: Remove unwanted background noise from field recordings or live performances to produce clean tracks.
- Dialogue extraction: Isolate speech from a video file using visual prompts to get the exact spoken parts.
- Instrument isolation: Study individual parts of a song by isolating the drums, bass, or vocals for practice or analysis.
How does SAM Audio work?
- Upload your audio file (MP3, WAV, FLAC, or other common formats) to the SAM Audio playground.
- Describe what you want to separate in natural language, or use a visual cue from a video or a time range.
- Select a separation mode: Everything, General Sounds, Music, or Speech for optimized results.
- Wait a few seconds for cloud processing, then download the separated tracks.
Pricing
SAM Audio offers free basic usage with no signup required. Additional features, higher processing limits, or priority access may be available via a premium subscription.
FAQ
What is SAM Audio?
SAM Audio is an online platform that provides access to Meta's SAM Audio AI model for audio separation. It allows users to isolate specific sounds from audio files using text, visual, or time-based prompts.
Is SAM Audio free to use?
Yes, basic usage is free and does not require an account. You can upload and separate audio directly on the website. Premium features may require a subscription.
What audio formats does SAM Audio support?
SAM Audio supports MP3, WAV, FLAC, and most common audio formats. You can upload files directly in the browser.
Can SAM Audio separate vocals from any song?
SAM Audio can separate vocals from a wide range of songs with high accuracy. Performance may vary depending on the complexity of the audio, but it handles most music tracks effectively.
Is SAM Audio the official Meta product?
No, SAM Audio is an independent platform that uses Meta's open-source SAM Audio AI model. It is not an official Meta product but benefits from the model's state-of-the-art capabilities.
