Isolate audio (isolate sounds): what it means
To isolate audio is to extract a target sound—speech, an instrument, an effect, ambience—from a mixture. In practice you may also see this described as separate audio or extract sounds.
Want the quickest result? Use the official demo, then refine your prompt until the isolated track is clean enough for editing.
Common workflows for isolating sounds
- Speech isolation: separate voice from background music.
- Sound effects extraction: isolate footsteps, claps, engine, birds, crowd noise.
- Music editing: isolate a specific instrument or a transient (e.g., snare hits).
- Video post: extract sounds from a noisy scene to rebuild a mix.
Tips: make the target sound more separable
- Use short clips where the target is loudest and clearest.
- Prefer nouns over adjectives: “applause” beats “happy background”.
- If available, add a span prompt to constrain the time range.
- Separate one sound at a time, then layer results in your editor.
Fast helpers
If you want copy-paste prompts, see prompt packs. If your output is messy, use troubleshooting.
Alternatives (classic source separation)
If you only need fixed stems (like vocals vs instrumental), classic tools may be enough. Many people compare against Demucs, Spleeter, or Open-Unmix for baseline separation workflows.
Back to home.