Remove vocals from a song online free and separate any song into Vocals, Drums, Bass and Other with the real Demucs AI model — running entirely in your browser. Use it as a free vocal remover, a free audio stem separator, a free karaoke maker or a free acapella extractor, then transpose chords online free with the built-in transposer. 100% Totally Free — No Registration, No Account, No Upload, No Watermark, No Download Limit on the tool itself. Your song is never uploaded: the AI runs on your own device, with no software to download.
Upload or drag & drop your audio file to get started.Upload File
Supports MP3, WAV, M4A and OGG • Your file stays on your device — nothing is uploaded
—
✓100% Totally Free✓No Registration✓No Upload — Runs in Your Browser✓No Watermark✓Unlimited Splits
Free vocal remover — real AI, not a filter trick
This page loads the genuine HTDemucs neural network (ONNX, 166 MB) and separates your song into four real stems on your device. No centre-channel tricks dressed up as “AI stems” — the actual model, the actual separation.
Free stem splitter no account needed — truly free
No trial, no premium tier, no sign-up form and no email address:
Free vocal remover no watermark downloads
Free karaoke maker no signup needed
Free stem splitter no account or registration
No credit system and no per-song fee
Every stem, the karaoke instrumental, all WAV downloads and the full chord transposer are free for everyone.
Private by design — no upload, no download
Most stem splitters upload your song to their server. This one cannot — there is no upload in the code path at all, and there is no app to install. Unreleased demos and personal recordings stay on your device.
AdvertisementAd space reserved for AdSense — Leaderboard (responsive)
100% Totally Free$0 forever — no trial, no premium tier
No RegistrationNo account, no email, no password
Free Vocal Remover No WatermarkNo upload — real Demucs AI, 4 stems on your device
Unlimited SplitsNo monthly limit, no cap, no download limit
Free Audio Stem Separator & Vocal Remover — Before You Start
Supports MP3, WAV, M4A and OGG. This free stem splitter (no account needed) decodes your file in your browser and the Demucs AI separates vocals, drums, bass free online into Vocals, Drums, Bass and Other — so you can remove the vocals, keep an acapella, or build a karaoke backing track. Nothing is uploaded and there is nothing to download or install.
In terms of use it is a free stem splitter no limit tool — unlimited splits, no monthly limit and no cap on how many songs you separate. Free use here means no monthly cap, no account, no watermark and no credits.
Honest technical limits — please read before your first song: These are device and model limits, not usage caps — there is no monthly limit, no credit system and no per-song fee on the tool itself.
First use downloads the 166 MB AI model once and your browser caches it, so later visits skip the download. Use Wi-Fi for the first song if mobile data is limited.
Chrome or Edge with WebGPU is best. Other browsers use the WebAssembly fallback and are slower.
Phones are slow and may run low on memory on long songs — a computer is strongly recommended.
A 3–4 minute song takes several minutes to separate. Keep this tab open and in the foreground while it works.
Separation is very good, not perfect. Demucs is a top open model, but faint traces of other parts can remain in a stem, especially on dense mixes.
2. Your Stems — Mix, Solo, Mute & Download Free, No Watermark
Press Play to hear all stems together in sync. Use Solo to hear one stem on its own, Mute to remove it from the mix, and the slider to set its level. Every stem downloads as a 44.1 kHz stereo WAV, free.
0:00 / 0:00
Instrumental & karaoke mix — free instrumental maker: one WAV combining Drums + Bass + Other — the full backing track with the vocals removed, ready to sing over.
Separate Vocals, Drums, Bass Free Online — The Four Stems You Get
Vocals
The singing and lead vocal line, separated by the AI — download it on its own with the free acapella extractor workflow, for practice, study, remixes or checking a melody.
Drums
Kick, snare, hats and percussion as one stem — useful for drummers learning parts and producers studying grooves.
Bass
The bass line on its own, so bassists can hear exactly what was played without the rest of the band on top.
Other
Everything else — guitars, keys, synths and strings — the harmonic body of the song in a single stem.
Isolate Vocals Online Free — Or Remove Them Completely
One separation, two directions. Download the Vocals stem alone when you want to isolate vocals online free for an acapella, a remix or study. Or download the Instrumental mix when you want the voice gone entirely for karaoke. It is the remove vocals from song online free no download workflow: no app to install, no file uploaded — the Demucs model runs in this page.
AdvertisementAd space reserved for AdSense — In-content Rectangle (responsive)
Paste your chords — or a full chart with lyrics — then transpose with the + / − controls. This chord transposer online free no signup tool changes only the chord lines; your lyrics stay exactly as they are, and the capo helper updates automatically.
Transpose Chords Online Free No Hidden Limits
Transpose chords online free no hidden limits, no credits and no account — use it for as many songs as you like, and copy or download every result free.
TRANSPOSE
Move up or down by semitones
0
Try an example:
Capo Helper
Change the transposition and we will show you exactly where to place your capo.
Quick Chord Finder
See what a single chord becomes in the new key — type any chord below.
Frequently Asked Questions
Is this free vocal remover and stem splitter really 100% free with no signup?
Yes — 100% totally free. This free vocal remover no signup tool lets you remove vocals, separate stems, download every stem and the instrumental mix, and use the full chord transposer online free. There is no trial, no premium tier, no credit system, no per-song fee, no registration, no account and no watermark ever on your downloads — free vocal remover no watermark, free karaoke maker no watermark, unlimited splits with no monthly limit and no cap. That is unlike Moises, which allows 5 tracks per month on its free tier, and LALAL.AI, which gives 10 free minutes as a preview only before payment.
How do I remove vocals from a song online free and separate vocals, drums and bass?
Add an MP3, WAV, M4A or OGG file above and press Separate. The page loads the HTDemucs neural network (the htdemucs_fp16weights ONNX model, about 166 MB) into your browser using ONNX Runtime Web. Your song is resampled to 44.1 kHz stereo and processed in 7.8-second overlapping segments, and the model outputs four stems: vocals, drums, bass and other. Everything runs on your own device — WebGPU if your browser supports it, with a WebAssembly fallback.
Can I remove vocals from a song online free with no download and no upload?
Yes — there is no upload and no software to download. Your audio file never leaves your device. Decoding, AI separation and WAV export all happen locally in your browser, and nothing about your audio is stored or transmitted by this site. Open the page, add your file, separate, and download the WAVs.
Is there a free acapella extractor to isolate vocals only?
Yes. Download the Vocals stem on its own as a 44.1 kHz WAV and you have an isolated acapella. The free acapella extractor is the same Demucs separation — just use the vocals stem instead of the instrumental.
What is the difference between a vocal remover, a stem splitter and a karaoke maker?
They use the same separation. A vocal remover takes the voice out (and gives you the voice on its own), a stem splitter separates vocals, drums, bass and other into four stems, and a karaoke maker gives you the instrumental backing track to sing over. This page does all three from one separation, plus a chord transposer online free.
Why does the first separation take longer?
On first use the browser downloads the 166 MB AI model once and caches it, so later visits start much faster. Separation itself also takes time: a typical 3–4 minute song needs several minutes of processing, and phones are considerably slower than a computer with WebGPU.
What quality should I expect from the separated stems?
Demucs is one of the best-known open stem-separation models and the results are very good for practice, covers, remixes and learning parts. No AI separation is perfect: you may hear faint traces of other instruments in a stem, especially on dense or heavily compressed recordings.
Which browser works best?
Chrome or Edge on a computer, with WebGPU enabled, gives the fastest separation. Other modern browsers fall back to WebAssembly, which works but is slower. Phones can separate short songs but are slow and may run low on memory with long tracks.
How does the chord transposer online free work — how do I transpose chords online free?
Paste chords or a full chart with lyrics, then use the + / − controls or choose an original key and a new key. Chords transpose instantly, lyrics stay untouched, and the capo helper shows where to place a capo to keep familiar chord shapes.
How do I use this as a free karaoke maker and free instrumental maker with no watermark?
Yes — this is a free karaoke maker no watermark tool and a free instrumental maker in one. After separation, the Instrumental download combines the drums, bass and other stems into one 44.1 kHz stereo WAV — a clean backing track with the vocals removed, ready for karaoke, covers and singing practice, with no watermark.
How to Remove Vocals From a Song Online Free — And Separate the Stems
Separating a finished recording back into its parts used to need the original studio multitracks. With this free audio stem separator it takes five ordinary steps, and none of them involves uploading your music anywhere or creating an account:
1Choose your song
Drag in an MP3, WAV, M4A or OGG file. A clean, good-quality recording separates better than a heavily compressed one, and a stereo mix gives the model more to work with than a mono file.
2Let the model load
The first time only, your browser downloads the 166 MB Demucs model and caches it. Watch the model progress bar — when it finishes, separation starts automatically, and future visits skip this step.
3Wait for separation
The AI works through your song in short overlapping sections. A typical song takes several minutes on a computer. Keep the tab open and in the foreground so the browser does not throttle it.
4Check each stem in the mixer
Play all four stems together, then Solo each one to hear it alone. If a vocal trace remains in the drums or the bass sounds thin, that is the honest reality of AI separation on that particular mix — dense, loud recordings are the hardest.
5Download what you need
Download individual stems as WAV files, save the Vocals stem alone as an acapella, or grab the Instrumental mix (drums + bass + other) for an instant karaoke backing track from the free karaoke maker. WAV is uncompressed, so the files are large but keep full quality for your DAW.
What Is a Free Stem Splitter? Stem Separation Explained
A “stem” is one part of a finished song on its own — the vocal without the band, the drums without the guitars. When a song is recorded in a studio, every part starts on its own track, but by the time you hear it as an MP3 those parts have been mixed down into a single stereo file. Separating stems means undoing that mix.
Older tricks tried to undo it with simple maths, like subtracting the left channel from the right to remove whatever sat in the middle. That sometimes quietens a vocal, but it also damages the music and cannot isolate drums or bass at all. Modern stem separation works differently: a neural network such as Demucs is trained on thousands of songs where both the finished mix and the true individual parts are known, so it learns what vocals, drums and bass actually sound like inside a mix. When you give it a new song, it predicts each part. That is why the results are far cleaner — and also why they are still predictions, not the original studio tracks. A Demucs stem is the best practical way to study, practise with, or rework a recording you do not have the multitracks for, whether you are a singer preparing a cover, a drummer learning a groove, a teacher making practice material, or a producer sketching a remix idea.
How to Transpose Chords Online Free to a New Key
Every singer has a comfortable range, and every song was recorded in exactly one key — often not yours. Transposing means shifting every chord in a song by the same number of semitones so the song sits where your voice (or your instrument) is happiest, without changing how the song feels.
Find the original key. If you know the song starts on C and the chords are C, Am, F and G, the original key is C major.
Decide how far to move. Sing the chorus. Too high and strained? Move down 1–2 semitones. Too low and murky? Move up 1–2. Small moves make a big difference.
Paste, transpose, check. Paste the chart into the transposer, set the amount, and read the new chords. Lyrics stay put — only chords change, with spacing preserved so the chart still lines up.
Use the capo helper. If the new key lands on awkward shapes like Eb or Ab, the capo helper shows you how to keep your easy open chords and let the capo do the work instead.
A practical example: a song in G major (G, Em, C, D) transposed down 2 semitones becomes F major (F, Dm, Bb, C). Same song, same feel — just kinder to a lower voice.
How to Make Karaoke for Free — No Watermark, No Signup
A karaoke track is simply the song with the lead vocal taken out, so your voice can take its place. With the free karaoke maker above, you separate the song once, then download the Instrumental mix — drums, bass and other together, with no singing on top. Because the separation uses Demucs AI rather than an old centre-cancel trick, the backing keeps its bass and drums at full strength instead of sounding thin. There is no watermark on the file, no signup before the download button, and no per-song fee: it is a free karaoke maker no watermark singers can use for practice, covers, parties and lessons. If you also need the song in a lower or higher key to suit your voice, run the chords through the chord transposer online free on this same page and play or sing along in the new key.
Why This Free Stem Splitter Is Different — No Signup, No Upload, No Watermark
Most “free” tools are free only until the moment you need the result. LALAL.AI, for example, gives 10 free minutes and lets you listen to a preview before paying to download in full, and Moises gives a small number of free tracks per month (5 on its free tier) before a subscription. This site is different by design: the tool itself has no minute cap, no monthly track cap and no preview-only wall. Here is the honest comparison:
Feature
This Site
Most Other Online Tools
Price
Totally free, forever
Free trial, then a subscription
Registration
None — use it instantly
Email account usually required
Your song’s privacy
Never uploaded — AI runs in your browser
Uploaded to and processed on their server
Separation engine
Real Demucs AI model, named openly
Undisclosed “AI”, quality varies
Watermark / limits
No watermark, no download limit on the tool itself
Watermarks, minute caps or preview-only downloads are common
Karaoke & acapella
Free karaoke maker + free acapella extractor included
No credit system, no per-song fee, no watermark ever
Credit systems and per-song or subscription fees are common
Signup to download
Free vocal remover no signup • free karaoke maker no signup • free stem splitter no account
Account often required before download
Chord transposer included
Yes — with capo helper
Usually a separate tool
About Music Stems and Chords
This little site started with a simple frustration. You hear a song, you want to sing over it without the original singer, or you want to hear exactly what the bass player did in the second verse — and every tool you find wants your email address first, your money second, and uploads your song to a server you know nothing about third. That felt wrong for something a computer can now do on its own.
So this site does it the other way round. It brings the AI to your music instead of sending your music to the AI. When you separate a song here, the Demucs model downloads into your browser once and then runs on your own machine. Your audio is decoded, separated and turned back into WAV files entirely on your device. It is never uploaded, never stored, and never listened to by anyone here — by design, not just by promise. The chord transposer works the same way: a small piece of music maths that runs instantly in the page, the way it should.
And it is free — properly free. No registration, no accounts, no trial clock ticking down, no “Pro” version holding your downloads hostage. The site costs very little to keep online because there is no server farm doing the heavy lifting; your browser does the work. In future the site may show advertising to cover its costs and keep the tools free, and if that happens it will be ordinary, clearly placed advertising — nothing that gets between you and the music.
Who is it for? Singers making backing tracks, guitarists moving songs into singable keys, drummers and bassists learning parts by ear, teachers preparing practice material, and anyone curious about what is actually going on inside a favourite recording. If it helps you make music, it is doing its job.
Contact Us
Questions, bug reports, or a song that separated badly and you want a second pair of eyes on the settings? We read everything. Because this site has no accounts and no server that receives messages, the contact form below does not send anything anywhere by itself — it prepares your message so you can copy it and send it through your own email or social app, or simply keep it as a note. When the site moves to its own domain, a direct contact address will be published on this page.
Tip: telling us your browser (for example, Chrome on Windows), the song length, and which step went wrong helps us help you faster. Please do not include passwords or payment details in any message — we will never ask for them.
Privacy Policy
Last updated: 6 November 2026. This policy explains, in plain language, what happens to your information when you use Music Stems and Chords.
Your audio stays on your device
All audio processing on this site — decoding your file, AI stem separation, mixing and WAV export — happens locally in your web browser. Your audio files are never uploaded to any server, never stored by us, and never shared with anyone. When you close or refresh the page, the processed audio held in your browser’s memory is gone. The AI model file (about 166 MB) is downloaded from its public host (Hugging Face) into your browser’s cache so the tool can run; that download contains no information about you or your music, and your browser controls its cache — you can clear it at any time in your browser settings.
No accounts, no personal profiles
There is no sign-up or registration on this site. We do not ask for your name, email address, or any account details to use the tools, and we do not build profiles of visitors. The optional contact form on this page does not transmit anything: it only prepares text on your device for you to copy.
Cookies, analytics and advertising
This page itself does not set tracking cookies. The page loads fonts and the ONNX Runtime library from public content-delivery networks (Google Fonts and jsDelivr), and those providers may log standard technical request data such as IP addresses under their own privacy policies. If, in future, this site uses analytics or advertising services such as Google Analytics or Google AdSense after moving to its own domain, those services may use cookies to measure visits and to serve and personalise ads, and this policy will be updated to name them and link to how you can opt out. You can block or clear cookies at any time in your browser settings; the tools on this page will still work.
Your choices
Because we hold no account and no uploaded content, there is nothing stored about you for us to export or delete. If you have a privacy question, reach us through the Contact section above.
Disclaimer
The tools on this site are provided free of charge, as they are, for personal, educational and creative use. AI stem separation produces an estimate of a song’s parts, not the original studio recordings: quality varies with the source recording, and faint traces of other instruments may remain in a stem. We cannot guarantee that separation will succeed for every file, browser or device, and processing happens on your own hardware, so results and speed depend on it.
Copyright: Songs you process may be protected by copyright. Separating a recording does not give you rights over it. You are responsible for how you use the stems you create — personal practice, learning and private enjoyment are the intended uses; publishing, selling or distributing material derived from recordings you do not own or have permission to use may infringe the rights of the artists, songwriters and labels involved. If you are a rights holder with a concern about this site, please reach us through the Contact section.
Nothing on this site is professional, legal or financial advice. External names and projects mentioned (such as Demucs, ONNX Runtime and Hugging Face) belong to their respective owners; mentioning them describes the technology used and does not imply endorsement.
Terms of Use
Last updated: 6 November 2026. By using Music Stems and Chords, you agree to these simple terms.
The service: a free, browser-based AI stem separator (vocals, drums, bass, other) and chord transposer, provided without registration. The service is offered “as is”, may change or be unavailable at times, and carries no guarantee of uninterrupted operation.
Your content: you keep all rights in the audio you process. Because processing happens entirely in your browser, we never receive, store or claim any rights over your files or the stems you create.
Acceptable use: use the tools only with recordings you own or are permitted to use, and only in ways that respect copyright and other applicable laws. Do not use the site to infringe intellectual-property rights, and do not attempt to disrupt, overload or misuse the site or the public services it loads its model and libraries from.
Liability: to the extent permitted by law, we are not liable for losses arising from use of the tools — including lost files, unsatisfactory separation results, or how separated material is later used. Always keep your own copy of your original recordings.
Changes: these terms may be updated; the current version will always be the one published on this page.