# AI sound remover and sound isolator for audio and video

Type any sound, such as dog barking, footsteps, keyboard typing, or drums, to remove it from your mix or isolate it on its own with Meta SAM Audio.

## How it works

1. Upload an audio or video file up to 10 GB (MP4, MOV, MKV, WebM, AVI, MP3, WAV, M4A, AAC, FLAC, OGG, Opus, or AIFF).
2. Choose Remove sound or Isolate sound under Mode, then describe the target sound in a short noun or action phrase—such as dog barking, piano playing, footsteps, or woman speaking.
3. Click Separate sound, compare the result with your original in the browser player, and download both tracks (with your original video picture preserved on the primary output).

## What you can separate with Audo's AI sound remover

Remove unwanted sounds from video takes, pull individual instruments from a song, or isolate clean foley effects from a busy recording:

### Video & dialogue cleanup: Remove dog barking, footsteps, typing, or door slams from video

Standard noise filters turn down steady hiss and hum, leaving sudden sounds like a dog barking mid-sentence, footsteps on a wooden floor, keyboard clicks, coughing, or a phone ringing right in the recording. Use Audo as an MP4 sound remover and video background sound remover: type the sound in a few words and Audo removes it from the soundtrack while copying your 1080p or 4K video stream without re-encoding. For steady room hiss or AC hum behind speech, use Remove noise, or fix room echo in Audio enhancer.

### Instrument & drum remover: Remove or isolate drums, guitar, piano, or solo instruments

Four-stem splitters group guitar, piano, strings, brass, and synths together into one Other track. With Audo's prompt-guided instrument remover and drum remover, set Mode to Isolate sound or Remove sound and type drums, acoustic guitar, piano, or saxophone to extract a part for ear training and sampling or make a custom practice backing track. When you want a song split into vocals, drums, bass, and other in one pass, use Stem splitter.

### Sound effect & crowd isolation: Separate sound effects, applause, laughter, or bells from recordings

Use Audo as a sound effect remover and audio isolator on field recordings, live events, and broadcast clips: strip audience applause, a laugh track, church bells, or wind gusts while keeping the surrounding scene intact, or pull a clean sound effect out on its own. To remove an entire background song under spoken dialogue, use Remove music from video; to strip everything except a speaking voice, use Voice isolator.

### Meta SAM Audio & API: Run Meta SAM Audio online in your browser, from code, or via AI agents

Running Meta SAM Audio (Segment Anything Model for Audio) yourself requires gated Hugging Face checkpoints, Python 3.11, and a CUDA GPU, while desktop wrappers require installing an app. Audo hosts Meta SAM Audio as an instant browser audio separator for files up to 10 GB and exposes separate_sound over Audo's REST API and MCP server so ChatGPT, Claude, Claude Code, and Cursor can remove or isolate sounds automatically—see Audio and video tools for AI agents.

## Text-prompted audio separation with Meta SAM Audio

Audo's online AI sound remover, audio sound remover, and audio separator extract or remove any sound you describe in plain English across audio (WAV, MP3, M4A, AAC, FLAC, OGG, Opus, WebM, AIFF) and video (MP4, MOV, MKV, WebM, AVI) files up to 10 GB using Meta SAM Audio (Segment Anything Model for Audio). Every run produces two time-aligned 48 kHz tracks: your recording with the target sound removed and the isolated sound on its own, while copying the original video picture untouched on MP4, MOV, MKV, WebM, and AVI uploads. You can separate the first 20 seconds of each file free, 3 times a day, with no sign-up, and process full recordings pay-as-you-go at $0.25 a minute ($15.00 an hour) from $5 with no subscription.

Traditional noise reduction attenuates stationary background hum, and fixed stem splitters only separate predefined categories such as vocals, drums, and bass. Meta SAM Audio pairs a T5 text encoder and Perception Encoder Audiovisual (PE-AV) alignment with a flow-matching Diffusion Transformer in a 48 kHz DAC-VAE latent space to separate open-vocabulary sounds across speech, music, and everyday sound effects. Because SAM Audio was trained on concise noun and verb-phrase captions, short 1-to-3-word phrases that name the sound or action—such as dog barking, piano playing, woman speaking, car honking, or footsteps—work markedly better than full sentences or instructions containing words like remove or without.

Because every run exports both files, you can open either track straight in Audo's online video and audio editor, or load both tracks into Mix audio to lower an intrusive sound by 10 dB instead of removing it completely. Clean up spoken dialogue next in Audio enhancer, trim isolated samples in Cut audio, or generate speaker-labeled transcripts in Transcribe.

How Audo's AI sound remover compares on free access, pricing, file limits, and video support (checked November 2026)

| Tool & model | Free tier | Paid pricing | Media & file limits | API & AI agent access | Source |
|---|---|---|---|---|---|
| Audo AI (/sound-remover · Meta SAM Audio) | First 20 s per file, 3 times/day, no sign-up, no watermark | Pay once from $5 ($0.25/min, charged only on success); no subscription | Browser app; audio & video up to 10 GB; copies original 1080p/4K video picture untouched | REST API, no-key Free API, and MCP server (https://audo.ai/mcp) | https://audo.ai/sound-remover?utm_source=markdown&utm_medium=agent |
| SoundScrub (Meta SAM Audio) | No browser tool (desktop download required) | $0.25 per 30 seconds ($0.50/min) prepaid packs; no subscription | Desktop Electron app (macOS, Windows, Linux); custom text prompts capped at 3 words | No public API or MCP server | https://soundscrub.video/ |
| EraseSound | 30-second preview (up to 20 previews per 24 hours, no sign-up) | From $4.99 for 10 minutes (~$0.50/min) or monthly plans | Browser app; MP3, WAV, M4A, MP4, MOV capped at 10 minutes and 1 GB | No public API or MCP server | https://erasesound.app/ |
| Sievin (Meta SAM-Audio) | 3-minute free trial | Starter $11.99/month (60 min/mo); Creator $38.99/month (200 min/mo) | Browser app; audio only (MP3, WAV, FLAC, M4A — no video) capped at 10 minutes | No public API or MCP server | https://sievin.com/ |
| Meta Segment Anything Playground | Free interactive research demo | No managed commercial tier or production SLA | Research playground; self-hosting requires gated Hugging Face checkpoints and a CUDA GPU | Open-weights Python code only (no hosted API) | https://ai.meta.com/research/samaudio/ |

Tool: Separate sound (`separate_sound`). Price: 25 credits a minute — $0.25/min ($15.00/hr), pay-as-you-go with no monthly subscription — of input for whole files, charged only when a job succeeds.

## Free API

```bash
curl -F file=@interview.mp4 -F 'sound=dog barking' https://audo.ai/api/try/separate-sound -D response.headers -o response.body
```

Read response.headers and response.body before naming media. Handle 202, 200 JSON and errors with the complete response client at https://audo.ai/docs/api?utm_source=markdown&utm_medium=agent#responses.

Free for the first 20 seconds of each file, 3 times a day per person, with no key or account. Uploads up to 10 GB, 2 at once from one address. The response is the result itself; a job that runs longer than about 30 seconds returns 202 with a link to check on it. With an API key (`-H "Authorization: Bearer $AUDO_API_KEY"`), omitting free runs the full job with credits if the balance covers it, otherwise the free allowance if quota remains. Set free=true to prohibit spending; for paid runs use free=false with an approved positive max_credits. A free result cut short comes with finishUrl (and the Audo-Finish-Url header when the response is the file itself): give it to the person, who pays there (from $5, no sign-up), and the whole file runs; keep checking the status link, whose finishedBy names the whole-file job and its own status link. Details: https://audo.ai/docs/api.md

## Options

Form fields with the same names as the MCP tool's inputs:

- `sound`: Short English noun phrase describing the sound to remove or isolate, such as "dog barking", "footsteps", "keyboard typing", "applause", "drums", or "acoustic guitar". Up to 200 characters.
- `keep`: Primary output mode: everything_else (remove the target sound and keep the rest, default) or sound (isolate the target sound). Both tracks are always returned.
- `output_format`: The result's format: same (as the input; a video stays a video), wav, mp3, m4a, or flac. Default same.

## Languages

Write the target sound prompt in English, such as `dog barking`, `footsteps`, `drums`, or `acoustic guitar`. Spoken audio in the file can be in any language.

## From an assistant (MCP)

Add https://audo.ai/mcp and sign in (setup for each assistant: https://audo.ai/setup.md). The tool is `separate_sound`:

> Removes or isolates any specific sound described in a short English prompt, such as "dog barking", "footsteps", "keyboard typing", "applause", "drums", or "acoustic guitar", in audio or video using Meta SAM Audio. Always returns two 48 kHz tracks: the recording with that sound removed and the isolated sound; keep selects the primary output (everything_else to remove the sound, the default, or sound to isolate it). On video with output_format set to same, the primary output copies the original video picture. For standard song stems (vocals, drums, bass, other) or dialogue/music splits, use split_audio. Costs 25 credits a minute, charged only on success; free for the first 20 seconds, 3 times a day. Returns download links, or job_id for get_job.

Example: "Use Audo to remove the dog barking from interview.mp4."

## Questions

**Is there a free AI sound remover and sound isolator online with no sign-up?** Yes. Audo separates the first 20 seconds of any audio or video file free, 3 times a day, with no sign-up and no watermark, giving you both the cleaned mix and the isolated sound. Full-length files cost $0.25 a minute ($15.00 an hour), charged only if separation succeeds, with pay-as-you-go packs from $5 that never expire and no monthly subscription—see pricing.

**How should I write the prompt, and which sounds can Meta SAM Audio detect?** Meta SAM Audio was trained on concise English noun and action phrases (noun + verb-ing), so describe the sound event itself rather than giving an editing command: write dog barking, car honking, piano playing, acoustic guitar, footsteps, applause, thunder, singing voice, woman speaking, or child voice—never remove the dog barking or a dog can be heard in the background (Mode already chooses Remove sound vs. Isolate sound). Keep prompts in English even if the speech in your file is in another language. When two sounds overlap closely in frequency or share the exact same voice type—such as two men speaking at the same time, a singer inside a choir, or rhythm guitar over bass—a faint trace can remain (use Speaker diarization to label speakers, or Bleep censor to mute specific words).

**Can I remove one sound from an MP4 or MOV video without muting the whole video?** Yes. Upload an MP4, MOV, MKV, WebM, or AVI file up to 10 GB on Audo and leave Save as on Same as the original. With Mode set to Remove sound, your primary download is the original video with its picture copied frame for frame and only the target sound removed, plus the isolated sound as a second audio file. If you want to silence the entire soundtrack instead, use Mute video.

**When should I use Sound remover instead of Remove noise, Voice isolator, or Stem splitter?** Use Sound remover when you want to remove or isolate one specific sound while leaving the rest of the scene intact, such as taking dog barks out of an outdoor video while keeping voices and ambient sound. When you only want clean spoken dialogue and everything behind the speaker can go, Voice isolator and Remove music from video are faster and cost less ($0.05 to $0.08 a minute). For splitting songs into vocals, drums, bass, and other in 44.1 kHz stereo, use Vocal remover or Stem splitter ($0.08 a minute).

**Does Audo run Meta SAM Audio, and what files do I get back?** Yes. Audo runs Meta SAM Audio (Segment Anything Model for Audio) in the browser with no Hugging Face checkpoint request, Python setup, or local GPU required. Every run returns two 48 kHz files named after your prompt: your recording without the target sound and the target sound on its own, saved in your original format or as WAV, MP3, FLAC, M4A, OGG, or Opus.

**How does Audo compare with SoundScrub, EraseSound, Sievin, and Adobe Audition Sound Remover?** SoundScrub requires installing a desktop app, limits custom prompts to 3 words, and charges $0.25 per 30 seconds ($0.50 a minute, twice Audo's rate). EraseSound caps files at 10 minutes and 1 GB at about $0.50 a minute ($4.99 for 10 minutes), while Sievin requires an $11.99 to $38.99 monthly subscription and accepts audio files only up to 10 minutes. Adobe Audition's Sound Remover requires a Creative Cloud subscription and manually highlighting a clean noise print on a spectral display. Audo runs in your browser on audio and video up to 10 GB for $0.25 a minute pay-as-you-go from $5, with free 20-second runs 3 times a day and no sign-up.

**Can ChatGPT, Claude, or coding agents remove or isolate a sound automatically with Audo?** ChatGPT and Claude cannot natively separate a specific sound from an audio or video file on their own. Once you Add Audo to your assistant via Audo's MCP server (https://audo.ai/mcp), ChatGPT, Claude, Claude Code, Codex, and Cursor can call separate_sound with your file, a sound prompt like dog barking, and keep set to everything_else or sound, then return download links for both tracks, or call the REST API directly from code.

## More

- Every tool: https://audo.ai/llms.txt
- OpenAPI: https://audo.ai/api/openapi.json
- This page for people: https://audo.ai/sound-remover?utm_source=markdown&utm_medium=agent