StemSlice logoStemSlice

Split a song online

AI Song Splitter

Turn one finished song into vocals, drums, bass, and other without installing audio software. Choose the section you care about, hear a free preview, and unlock the whole song only when the result fits your project.

Split a SongHear a Real Demo
30MBfree upload
30 secfree preview
Shownfile deadline

Separation job

Outputs vocals, drums, bass, and other.

Secure upload → AI separation → Preview → Download. Free previews use up to 75 seconds around your selection; 1 credit processes the full track.

Compare previews0:00–0:30 preview
1

Add the song from your device

Choose MP3, WAV, FLAC, M4A, AAC, or OGG. Free uploads support 30MB and 6 minutes; the paid path supports 500MB and 20 minutes.

2

Pick the 30 seconds you want to judge

The free job processes up to 75 seconds around your selection, then lets you compare the original with vocals, drums, bass, and other.

3

Unlock the complete song if it works

One credit processes the full track and unlocks four core stems plus instrumental, drumless, and bassless mixes in your selected format.

Hear real separation examples

These recordings are licensed under CC BY 4.0. Compare each available result with its original at the same timestamp. User uploads remain private and are never added here.

One song, four parts you can actually use

Start with the musical section that matters, then keep the full aligned package only when the preview suits the project.

Plain-language song splitting

No model picker or desktop setup. Upload a song, audition its separated parts, and decide from the actual source you plan to use.

Useful for karaoke, practice, and remixing

Keep the vocal, remove it for a backing track, solo the rhythm section, or import all four aligned stems into an editor.

Four true stems

Vocals, drums, bass, and other instruments are exported as separate files you can drop into a DAW.

DAW-friendly formats

Paid downloads support WAV, FLAC, and 320kbps MP3 without watermarking. WAV and FLAC are workflow-friendly conversions from the separated MP3 audio.

Private by default

Uploads and generated results are encrypted in storage. The page shows the access deadline, including the later deadline granted after a paid result becomes ready; automated cleanup removes expired objects afterward.

No subscription required

Buy credits once and use them whenever. One complete 7-file package costs 1 credit.

Split a song without learning a production system

Use plain listening checks for karaoke, practice, arrangement study, and a first remix sketch.

Split a song when you need parts, not production jargon

A song splitter is the practical version of a stem-separation workflow: add one finished song and receive vocals, drums, bass, and the remaining instruments as aligned files. You do not need to know how a source-separation model is trained or how multitrack sessions are organized. The useful question is whether the separated parts help with the job in front of you. Start with a 30-second excerpt from a musically important section, switch between the original and each result, and only unlock the complete song when the preview answers that question.

Make a karaoke backing track without installing an editor

For a quick karaoke rehearsal, the instrumental result combines drums, bass, and other while leaving the predicted vocal out. Check the first entrance, a busy chorus, a quiet verse, and the ending after the full result is ready; those moments reveal different kinds of vocal residue. Keep in mind that backing vocals, vocal reverb, and instruments occupying the same frequencies can affect the result. A browser workflow is convenient for deciding whether the track is usable, while public performance, streaming, or commercial distribution still requires the appropriate permission for the underlying song and recording.

Turn the same split into focused practice material

Musicians can use one job in several ways. A bassist can solo the bass line, a drummer can compare the kit against the full arrangement, and a singer can lower the vocal while keeping the accompaniment intact. Because every result shares the same source start, the files remain aligned when imported together. Do not trim silence from one stem before the others or normalize every file independently; either change can make a correct split appear out of time or unbalanced. Preserve an untouched copy, then create practice versions for the instrument, section, and tempo you actually need.

Prepare remix ideas before committing to a full export

A remix rarely needs every possible edit on the first pass. Use the free preview to test the hook, drop, breakdown, or transition you expect to reuse. Listen for stable vocal consonants, drum transients, bass note definition, and instruments that leak between predictions. If the passage survives that check, one credit processes the complete track and unlocks four core stems plus instrumental, drumless, and bassless mixes. This approach keeps experimentation lightweight: reject an unsuitable source after the preview instead of spending time downloading and organizing a full package that will not enter the session.

Understand what the four song parts contain

Vocals can include lead, doubles, harmonies, breaths, and vocal effects. Drums generally contain the kit and much of its room sound. Bass targets the main low-frequency instrument, while other collects guitars, keyboards, strings, synths, effects, and anything not assigned to the first three groups. These are predictions from a mastered mix, not access to the original studio multitracks. Shared reverb or overlapping timbres may appear in more than one file. Naming the parts accurately makes it easier to decide whether to solo, mute, recombine, or edit them.

Choose an export format for the next device

MP3-320 is compact for rehearsal phones, messaging, and quick listening. WAV is widely accepted by editors and DAWs, while FLAC reduces storage size without adding another lossy encode. The separation service returns compressed model audio, so converting that result to WAV or FLAC improves workflow compatibility but cannot restore detail already absent from the prediction. Long songs also take more time and space to convert. Pick one format based on where the files go next, download the approved package once, and keep the ZIP with the source name and project notes.

Use source quality to set a realistic expectation

A clean studio master usually gives the model clearer boundaries than a clipped phone recording, a low-bitrate repost, or a live capture full of crowd noise. Heavy distortion, wide reverbs, layered vocals, and instruments sharing the same pitch range can still create audible overlap. That does not make every result unusable: light bleed may disappear inside a rehearsal mix while missing transients can matter greatly for sampling. Judge the exact section and intended use rather than treating one demo, genre, or marketing label as a guarantee for every song.

Preview the section that answers the buying decision

The free job processes up to 75 seconds around the excerpt you select and returns a 30-second listening window. Put that window over the chorus for karaoke, the groove for instrument practice, or the transition for a remix. A successful preview does not spend a credit. If it is useful, one credit starts full-track processing and unlocks the complete seven-file package; credits bought in one-time packs do not expire, and a failed full-track job automatically returns its reserved credit.

Review the complete song before a rehearsal or release

A 30-second excerpt is a purchase check, not proof that every verse and ending behaves identically. After full-track processing, listen across the complete arrangement and confirm that the seven exported files have matching duration and start time. The page displays the file deadline; source and free-preview files normally expire 24 hours after job creation, while a later paid result remains accessible until at least 24 hours after it becomes ready. Download the approved package before that deadline.

Credits that do not expire

1 credit unlocks 4 stems plus drumless, bassless, and instrumental mixes. Previewing, listening, and failed processing jobs do not consume credits.

Loading available credit packs…

FAQ

Answers for free limits, formats, privacy, copyright, and download quality.

How do I split a song into separate parts?

Upload a supported audio file, choose a 30-second section, and start the free preview. The result lets you compare vocals, drums, bass, and other at the same timestamp before deciding whether to process the whole song.

Can I use the song splitter for free?

Yes. The free path accepts files up to 30MB and 6 minutes and processes up to 75 seconds around your chosen excerpt to make a 30-second preview. It does not use a credit.

What does one credit unlock?

One credit processes the complete song and unlocks four core stems—vocals, drums, bass, and other—plus instrumental, drumless, and bassless mixes. If full-track processing fails, the reserved credit is returned automatically.

Can I make a karaoke track or practice mix?

Yes. Use the instrumental for singing, solo an instrument for transcription, or recombine aligned stems for a custom practice mix. Public or commercial use still requires the appropriate rights to the song and recording.

How long do credits and files last?

Credits from one-time packs do not expire. Audio files are temporary: source and free-preview files normally expire 24 hours after job creation, and a paid result remains available until at least 24 hours after it becomes ready. The job shows the exact deadline.

Related toolsStem SplitterKaraoke MakerAbout StemSlice

Extracted stems remain the property of the original rights holders. Commercial use requires their permission.