Split your song into four stems: vocals, drums, bass and other. Compare the parts, make a backing track and download every full WAV or one ZIP, free.
Hear a real four-stem exampleChecking free separation availability…
Drop your audio here
Vocals, drums, bass and other.
MP3 · WAV · FLAC · M4A · OGGUp to 10 minutes · 100 MB per fileFour stems · No sign-up · Up to 10 minutes
Free WAV downloads. Up to 5 jobs per browser per day, shared by the two-track and four-track tools. The allowance resets at 00:00 UTC. Site capacity and network protection can pause processing even when you have jobs left.
Previewing a file or listening to examples uses no jobs. Download your results within 55 minutes after processing finishes.
Compare the original with vocals, drums, bass and other from an actual Spleeter four-stem run. Switch tracks to stay at the same point in the recording.
Listen for what carries over. Check for kick or vocal residue in the bass, lost note attacks and remaining low notes in the backing. A separated stem is an estimate of an instrument group, not the original studio track.
The previews are 192 kbps MP3s for smaller downloads. Your own separation exports WAV stems. This is one short example, not a quality score or a promise for every song.
Slow Motion Dream by Steven M Bryant, © 2011, featuring CSoul, Alex Beroza & Robert Siekawitch. CC BY 3.0. 10.88-second excerpt from the Spleeter example. Changes: AI separation and MP3 preview encoding. No creator endorsement. Audio provenance.
LISTEN & LEARN · 6 min readHow to Practice Bass with Song StemsHear bass and vocal stems, follow a quieter guide, then take over the bass part.Drop a local MP3, WAV, FLAC, M4A or OGG file into the upload area, or select Choose audio file. Use a recording you own or have permission to process. Preview it on your device and check the filename, duration and size. The limit is 10 minutes and 100 MB per file; your browser checks the duration before uploading.
Confirm your audio rights, then select Split into 4 stems once. Your recording uploads to private temporary storage for AI stem separation. The upload bar shows actual transfer progress; processing then shows its current status. Keep this page open. If the connection drops, check the existing job instead of submitting the same recording again.
Preview your four stems and download the original WAVs or ZIP on this page. Open Adjust & practise when you want to change levels, mute or solo parts, loop a passage, or export a mix as MP3 or WAV. Save the audio within 55 minutes of completion; refreshing does not extend that window.
A finished song combines many sounds into one file. Stems make those sounds easier to work with in groups. This audio stem splitter produces four parts that share the recording’s timeline, so you can compare the same moment in each.
Lead singing, backing voices and vocal harmonies belong in the vocals stem. Use it to study phrasing, timing or the balance between lead and backing parts. Reverb and traces of instruments can remain. A double-tracked chorus stays grouped here; this is not a separate file for every singer or microphone.
Extract an acapellaThe drums stem brings together the rhythmic percussion: kick, snare, cymbals and related sounds. Listen for the groove, fills and accents, or practice a pattern slowly in your own player. Sharp guitar attacks and other percussive sounds can overlap with drums, so check the whole passage before sampling an isolated hit.
Extract the drum trackThe bass stem focuses on the bass line and low-end movement. It can help you hear note lengths, follow a progression or understand how the bass locks to the kick. Low piano notes, synths and drum energy may bleed into it. It is an estimated musical part, not a perfect frequency filter.
Extract the bass lineOther gathers the remaining accompaniment, often guitar, piano, keyboards, strings and sustained effects. Use it to study chord changes and arrangement space, or as a starting point for your own licensed edit. These instruments remain combined. Four stems describes the output groups; it does not mean every instrument is isolated individually.
The AI looks for patterns in a mixed recording that resemble voices, drums, bass and the remaining instruments. It estimates how much of each sound belongs in each group, then reconstructs separate stems. It works from the audio you upload, without access to the original recording session.
Sounds share frequencies and overlap in time, which makes the task imperfect. A cymbal and a breath may blend together; a reverberant vocal can spread through the accompaniment. Use this free AI stem splitter as a practical starting point, then listen and decide whether each result suits your project.
Bring the four WAVs into an audio editor to experiment with the balance of your own song. Keep the files aligned at the same start time, then adjust or replace a part. Listen to the stems together as well as solo: artifacts that stand out alone may sit differently in a full arrangement.
Choose Vocals and Listen to my part to study phrasing, then Play with the band for rehearsal. Adjust levels as you listen. For two tracks, the voice remover opens with Instrumental selected. Sharing a cover or edit may require additional permissions; separation does not supply those rights.
To study the voice before singing, use the vocal isolator. For an instrumental you can save and rehearse with, prepare a backing track.
Choose drums or bass: listen alone, use a quiet guide or play with the band. Name and loop passages, adjust speed and pitch, then export MP3 or WAV. Exports allow five minutes with changes or ten at original settings. Save your setup and original WAV stems to resume in Audio Practice.
Set your metronome with the automatic BPM finder, compare a file estimate with your taps in the BPM counter, or check the beat by ear with tap tempo.
For guided bass practice, open Adjust & practise in your results, choose Bass under I’m practising, then listen to bass, use a quiet guide or play without bass. The bass remover opens with the bass-free backing selected.
Use stem separation to explore how the voice, low end and percussion share space. Compare entrances, dynamics and arrangement decisions. The separated files are useful listening references, but they cannot reveal the original studio processing chain, microphone recordings or exact fader settings. Avoid treating them as the original multitrack session.
To hear how low-frequency EQ changes the full recording, compare the original with bass boost. This adjusts the mix on your device without separating it.
Up to 5 jobs per browser / day. No account, payment or watermark is required. The allowance is shared with the vocal isolator and resets at midnight UTC. Each job handles one file up to 10 minutes and 100 MB. Site capacity and separate network safeguards also apply, so an unused personal allowance does not guarantee an immediate processing slot.
Confirmed failed or canceled jobs return to your original browser’s allowance once. An uncertain submission stays reserved while its status is checked: it may still be running. Use Check existing job to recover it without starting another separation.
Keep all four full-length WAV files, individually or in one ZIP. Uploads are temporarily stored and deleted automatically. Download links expire 55 minutes after processing finishes. The upload retention period does not extend download availability. See the privacy policy for deletion timing. Save your results promptly; this anonymous tool does not provide an account history.
Bass practice includes three versions in MP3 or WAV. Preparing these mixes uses the same separated tracks and does not use another separation allowance. Practice preparation has its own limited capacity; your four original stems remain available if it is busy.
Choose MP3, WAV, FLAC, M4A or OGG from your device. Mono and stereo inputs are supported; the server checks the actual audio before processing. All four outputs are WAV. Converting a compressed source to WAV does not make the original recording lossless or improve its missing detail.
Start with your cleanest available recording. If a file is too long, trim it in an audio editor before selecting it here. If preview is unsupported on your phone, try MP3 or WAV. For large results, save tracks individually if your browser has trouble preparing the complete ZIP.
Clear sources with distinct parts are easier to separate than dense, distorted or heavily compressed mixes. Live-room sound, crowd noise, reverb and overlapping instruments can leave audible bleed. Quiet details may disappear or sound watery. These are limits to assess by listening, not a reason to assume every successful job is production-ready.
Compare a verse, chorus and busy transition with the original. Check the stem alone and in your intended mix. Separation does not restore clipping or reveal hidden studio tracks; choose the parts that are useful and keep the original as your reference.
Yes. You can use this AI stem splitter free of charge, including all four full-length WAVs and the ZIP, with no sign-up or payment. Up to 5 jobs per browser / day are shared across our separation pages and reset at midnight UTC, subject to site capacity. Each job accepts one file up to 10 minutes and 100 MB. Confirmed failed or canceled jobs restore the original browser allowance once.
Stems are grouped audio parts from a recording. You get four: vocals, drums, bass and other. Other includes the remaining accompaniment, such as guitar and piano. The AI estimates these groups from the mixed audio; it does not recover the original studio tracks or separate every individual instrument.
A stem splitter separates vocals, drums, bass and other accompaniment. A vocal isolator creates two tracks: vocals and a combined instrumental. For karaoke backing or an acapella, use the free vocal isolator. Choose four stems when you also want to work with the rhythm section separately.
Free vocal isolatorNo. This tool only accepts local audio files uploaded from your device. It does not import YouTube or other streaming links. Choose audio you own or have permission to process, and check your rights before sharing the resulting stems.
Upload MP3, WAV, FLAC, M4A or OGG, up to 10 minutes and 100 MB per file. Mono and stereo recordings are supported. The four outputs are WAV files. If your browser cannot preview a format, export a compatible MP3 or WAV from your audio editor and try that file.
Use them for practice, a cover, a remix of your own work or a mixing reference. Download each part separately or save all four in a ZIP, then open them in your preferred audio editor. Commercial use and distribution depend on your rights to the recording and underlying music.
An AI stem splitter estimates overlapping sources, so results vary. Dense arrangements, reverb, distortion and low-quality recordings can leave bleed, missing detail or watery artifacts. Listen to each stem and compare it with the original. WAV output avoids another lossy encoding step, but it cannot restore detail already missing from the source.