An acapella extractor separates the vocal part of an audio recording from its instrumental backing. Add an audio file above, leave Vocals selected, and the tool returns a vocal MP3 you can preview and save. The same separation also makes an instrumental mix and four underlying stems, so you can check what the model placed in each part. You do not need an account.
What you get from one upload
The default Vocals choice puts the voice first. It also offers the instrumental mix, which is the backing track left after the vocal stem is removed. If the goal is a sing-along track, you can start directly on the vocal remover page, where Instrumental is preselected. Both choices use the same audio-processing pass; they change which result the page puts in front of you.
Choose Drumless when you want the song without the drum stem. That result keeps the vocals, bass and other instruments. It can help you hear a rhythmic part more clearly while practicing, but a drum sound that overlaps another instrument may still be audible. Choose Four stems for the separate vocals, drums, bass and other parts. The stem splitter page explains what “other” means and how to use the four-part preview.
The files you download are 320 kbps MP3s. The original file you upload is not changed on your device. There is no WAV or FLAC output at launch, even if your input was WAV or FLAC. If you need a different format for a project, convert the MP3 in your audio editor after downloading it.
Getting a cleaner vocal track
Start with the clearest source you have permission to use. A low-bitrate copy that already sounds watery or distorted will give the separator less detail to work with. A stereo mix with a fairly clear lead vocal often gives the model a more obvious target than a recording where the voice is heavily layered into synths or guitars. That is guidance, not a guarantee: the sound of the original mix controls much of the result.
Listen to the first few seconds and to a busy section before you download. Reverb tails and backing vocals can land partly in the instrumental stem. Cymbals or bright instruments may land partly in the vocal stem because they share frequencies with the voice. If a passage matters for a remix, use the preview to find these spots and edit them after download. The tool gives you parts to work with; it does not replace detailed audio editing.
For a voice-only practice track, try the Vocals result first. For an instrumental backing, try the Instrumental result. If the lead is still too audible in the backing, compare it with the four-stem preview: muting Vocals there lets you hear what remains in drums, bass and other. The browser mixer only previews; it does not create an additional edited download.
Limits, privacy and rights
You can upload an MP3, WAV, FLAC, M4A or OGG file of up to 50 MB and 10 minutes. The browser checks the size and, when it can read the file, the duration before upload. The processor checks the duration again. A file that cannot be decoded may fail after upload. Processing time depends on length and server capacity, and the page shows an honest upload percentage followed by a processing status rather than a made-up completion percentage.
There are three free jobs per day per visitor/IP. The limit helps keep a shared GPU service available; people on the same network may share it. Temporary audio files are stored privately. Access expires after one day, and storage lifecycle deletes them afterward. The site does not keep a public collection of songs or stems. The privacy page describes the temporary files, job records and anonymous site statistics.
Upload audio you own or have permission to process. An extracted vocal is still part of the original recording. If you plan to publish a remix, mashup or sample, check the rights for that use before sharing it. For private practice, the same caution helps you avoid putting someone else’s recording where you did not intend. Our terms explain the upload rules in plain language.