By the end of this guide you will have split a song into separate tracks, such as a vocal-only file and an instrumental, and exported them as audio you can use for karaoke, practice, remixing or editing. An AI stem separation tool does the work: you upload the song, choose which parts to extract, preview the result and download the files.
If you want to separate vocals from a song with AI, the main choices are which web service or app to use, how many stems you need, and what your rights are over the source recording. This guide walks through the steps and compares seven tools from our Audio and voice category.
Key takeaways
- Upload a clean, high-quality copy of the song, choose the stems you want, preview a short section, run the full split, then check the files for leftover bleed before using them.
- The main decision is how many stems you need: two for karaoke or an acapella, more for remixing or practice. LALAL.AI, Remusic and Voice-Swap list different stem sets.
- Free allowances are small, for example 10 minutes on LALAL.AI's Starter plan, and some free tiers do not allow downloads.
- You need the right to use the recording. Separating a song does not give you permission to publish or sell the parts.
What you need before you start
- The song file. The highest-quality version you have, such as WAV, FLAC or a high-bitrate MP3. Low-quality files give muddier stems.
- The right to use it. Your own recording or one you have permission to remix. Check this before you publish anything.
- An account or an app. Check each service's sign-up terms. LALAL.AI also offers desktop and mobile apps and plugins for DAWs.
- Time. Under an hour for a first song, including a preview and a re-run.
- Difficulty. Easy. Most of these tools ask you to upload a file and pick the stems you want.
- Headphones. Bleed between stems is easy to miss on small speakers.
Which kind of AI tool fits this job
Pick by how many stems you need and where the processing should happen.
A dedicated stem splitter. LALAL.AI is built around AI stem separation: you pick which stems to extract (vocals, instrumental, drums, bass, guitar, synth, strings or wind), listen to a short preview, then run the full split. It runs on the web, in desktop and mobile apps, and as VST3 plugins for DAWs.
A vocal toolkit for musicians. Kits AI is a browser-based vocal production toolkit with a vocal remover and stem splitter, alongside voice conversion and cloning. Voice-Swap lists four-stem separation into vocals, drums, bass and other, and a free VST/AU plugin for use in a DAW.
An all-in-one music studio. Remusic has a vocal remover that splits a track into vocals, bass, drums, guitar and piano, plus a karaoke maker that removes the vocals, syncs lyrics and exports a karaoke video. AI Song Maker lists vocal removal and stem extraction within a song-generation tool.
A noise-cleaning tool with vocal separation. Noise Reducer is mainly for cleaning speech, but it also has a vocal separation tool that isolates vocals or instrumentals from songs. AIMusic is a song generator that also has a vocal remover and a stem splitter.
Step 1: Choose the song and confirm your rights
You end this step with a source file you are allowed to use.
- Pick the highest-quality file you have.
- Decide the purpose: karaoke, practice track, remix, sampling or cleaning dialogue.
- Confirm you own the recording or have permission. Splitting a commercial release does not change who owns it.
- Note whether the project is commercial, since that affects whether you can publish the results.
Check it worked by writing down the purpose and the rights you hold in one line.
Step 2: Decide which stems you need
Extracting only what you need keeps the job short and the allowance intact.
- For karaoke or an acapella, you need two stems: vocals and instrumental.
- For remixing or practice, you may want drums, bass, guitar or piano separately. LALAL.AI lists vocals, instrumental, drums, bass, guitar, synth, strings and wind, and Remusic lists vocals, bass, drums, guitar and piano.
- For a DAW workflow, check plugin support. LALAL.AI's VST plugin is limited to the Pro plan.
- Match the stems to the tool in the comparison table below.
Check it worked by listing the exact files you expect to download at the end.
Step 3: Choose a tool and check the limits
You end this step with a tool whose allowance fits your song.
- Check the free allowance. LALAL.AI's Starter plan is always free and includes 10 minutes with a 200MB upload limit per file. Kits AI's Free plan gives 15 conversion minutes with no downloads.
- Check whether you can download on the free tier. Remusic gifts some credits daily to try features, and its paid plans include downloads.
- If you work inside a DAW, look for plugins. LALAL.AI lists VST3 plugins, and Voice-Swap lists a free VST/AU plugin.
- Pick the export formats you need. LALAL.AI exports MP3, WAV, FLAC, OGG or AAC.
Check it worked by confirming your song length is inside the allowance.
Step 4: Upload the song and preview
The goal is a quick listen before you spend the allowance on the full track.
- Upload the song, or open it in the app or plugin.
- Select the stems to extract.
- Listen to the short preview if the tool offers one, as LALAL.AI does.
- If the preview has audible bleed, try a different neural network option. LALAL.AI lists a choice between neural networks on the same track.
Check it worked by hearing the vocal clearly with little backing in the preview.
Step 5: Run the full split
Now run the job on the whole song.
- Start the full-length process.
- Wait for each stem to finish. Hosted services use credits or minutes, so note what the job used.
- Open the results and play each stem on its own.
- For long songs, allow extra time and keep the page open until every stem finishes.
Check it worked by seeing every stem you chose listed and playable.
Step 6: Check for artefacts and clean up
AI separation is good but not perfect, so listen for faults.
- Listen to the vocal stem for backing instruments leaking in, and to the instrumental for ghost vocals.
- Listen for watery or metallic sounds on sustained notes and reverb tails.
- If the vocal stem is noisy, run it through a cleaner. LALAL.AI has a Voice Cleaner, and Noise Reducer removes hiss, hum and echo.
- Re-run with different settings if needed. LALAL.AI lists Enhanced Processing and De-Echo settings.
Check it worked by playing the vocal stem against the original and hearing the same performance.
Step 7: Export and use the stems
Save the files in a format that fits your project.
- Download each stem as WAV or FLAC for editing, or MP3 for sharing.
- Name files clearly, for example song-vocals and song-instrumental.
- For karaoke, Remusic's karaoke maker removes the vocals, syncs lyrics and exports a karaoke video.
- Keep the original, and do not publish or sell the stems unless you hold the rights.
Check it worked by importing the stems into your editor and playing them together at the original balance.
Tools for this job compared
| Tool | Good fit for | Pricing | Stated limits |
|---|---|---|---|
| LALAL.AI | Multi-stem splitting with preview (web, desktop, mobile, API) | Freemium, paid from $7.5/mo (checked 24 Oct 2026) | Starter: 10 minutes and a 200MB upload limit per file; VST plugin and API access are Pro only |
| AIMusic | Vocal remover and stem splitter beside a song generator (web) | Freemium, paid from $9.99/mo (checked 24 Oct 2026) | Free plan storage lasts 7 days; unused monthly credits do not roll over; one job at a time on free accounts |
| Kits AI | Vocal remover and stem splitter for musicians (web) | Freemium, paid from $10/mo (checked 24 Oct 2026) | Free plan has 0 download minutes; unlimited usage subject to fair use |
| Remusic | Stems and karaoke video in one studio (web) | Freemium, paid from $49/yr (checked 24 Oct 2026) | Vocal removal consumes credits; refunds on credit packs only within 12 hours if no credits used |
| Noise Reducer | Cleaning stems and basic vocal separation (web) | Freemium | Free files up to 20 minutes within 120 minutes per 24 hours; 500 MB audio upload limit |
| Voice-Swap | Four-stem splitting and voice models (web, VST/AU plugin, API) | Paid | Subscription credits reset monthly and do not roll over; plan prices not shown |
| AI Song Maker | Vocal removal alongside song generation (web) | Freemium, paid from $10.49/mo (checked 24 Oct 2026) | Free plan uploads of custom music limited to 1 minute; 20 credits per day |
Prices and plan limits change, so confirm the current terms on each vendor's site before you subscribe.
Common mistakes to avoid
- Starting from a low-quality file. Compressed audio gives muddier stems because detail is already gone. Fix it by using WAV, FLAC or the highest-bitrate copy you have.
- Using the full allowance on the first try. A wrong setting wastes minutes. Fix it by previewing first where the tool offers it.
- Publishing stems without rights. Separating a track does not transfer ownership. Fix it by using your own recordings or getting permission.
- Expecting a perfect split. Backing vocals, reverb and dense mixes leave artefacts. Fix it by listening on headphones and re-running with different options.
- Forgetting file-size caps. Free tiers limit upload size, as LALAL.AI's 200MB cap does. Fix it by checking size before you upload.
How we put this together
This is an evidence-based guide built from each tool's listing in our directory, which is compiled from the vendor's own site and pricing page. It is not a review based on using the tools. We don't rank tools, and nobody paid to be included. Some outbound links may be affiliate links; see our disclosure.
Frequently asked questions
Can I separate vocals from a song for free?
Yes, within limits. LALAL.AI has an always-free Starter plan with 10 minutes, AIMusic gives free accounts 15 credits a day, and Remusic gifts some credits daily. Free tiers can limit downloads, as Kits AI's Free plan does, so check what you can export.
Can I separate vocals from a song on my phone?
LALAL.AI lists iOS and Android apps. The other tools here are listed as web or plugin tools, so test them in your phone's browser before you rely on them, and check each vendor's site for current mobile features.
Is it legal to remove vocals from a song?
The law depends on your country and on what you do with the result. Private practice is different from publishing or selling. You need the rights to the recording to release stems or remixes, so ask the rights holder, and take legal advice if the use is commercial.
How many stems can I get?
It depends on the tool. LALAL.AI lists vocals, instrumental, drums, bass, guitar, synth, strings and wind, Remusic lists vocals, bass, drums, guitar and piano, and Voice-Swap lists four stems: vocals, drums, bass and other.
How long does it take?
The listings give no standard times, so run a short preview first, and leave time to check the stems.
Compare more options in our Audio and voice category, or see how to make a song with AI if you want to create new music instead.
