Bring your track
Choose an MP3, WAV, M4A, or FLAC you own or have permission to process. At launch: up to 10 minutes and 100 MB per song.
Get closer to the performance
Vocal Isolator & Acapella Extractor
Follow every phrase. Explore the vocal at the center of a song, from the first breath to the last harmony.
VocalsDrumsBassOther
A workspace for practice, remixes & new ideas.
Your next mix starts here
Drop your audio, or choose a file.
MP3, WAV, M4A, FLAC · Up to 100 MB / 10 min at launch
Choose a file to preview the workflow.
Preview only · No audio is uploaded
Listen a little closer
One track. Different perspectives.
ORIGINAL SESSION · 100 BPM
An original synthesized arrangement, played as a mix and its source parts. No sung vocals; this illustrates stems, not AI separation quality. Audio source & credits
The planned workflow
From a song to a starting point.
Find your parts
Choose vocals and instrumental, or vocals, drums, bass, and other instruments. Processing and progress tracking are coming next.
Listen, then decide
The planned free tier offers one short preview per day. Full-track WAV and MP3 downloads will be available with paid minutes.
A closer look
Vocal Isolator & Acapella Extractor: what to listen for
A vocal isolator estimates the singing in a mixed recording and separates it from the instrumental accompaniment. The vocal stem can help you study phrasing, work out harmonies, transcribe lyrics, or experiment with an acapella arrangement. Depending on the recording, the result may include lead vocals, doubles, backing singers, and vocal effects together. It is usually a vocal group rather than a separate file for each singer.
Listen closely to the beginning and end of each line. Breaths, consonants, and reverb tails reveal different weaknesses from sustained notes. Check whether the singer sounds natural when the instruments get louder, and listen for snare hits or chords leaking into the vocal. When testing harmonies, use a passage where the backing parts enter and leave; that makes it easier to hear whether quieter voices have been retained.
An isolated vocal can be useful for learning even when it is not clean enough for an exposed production. For a remix, place it into your arrangement before making a final judgment, because added instruments can hide some bleed while highlighting other timing or tonal issues. Match the tempo carefully and confirm that you have the rights to use the original performance. An extraction tool does not change who owns the recording.
Work from a good source
Keep a copy of the original recording, and avoid adding compression or effects before separation. Turning a low-quality file into a larger format will not restore detail. Use the same section when comparing outputs, and listen at similar volume. The most useful result is the one that serves your next step, whether that is learning a phrase, rehearsing with a backing track, or testing an arrangement in your audio editor.
Good questions
A few things to know.
Will lead and backing vocals be separate?
The planned two- and four-stem modes group vocals together. Separate lead and backing vocal extraction is not part of this launch scope.
Can I extract vocals from a live recording?
Live recordings can be challenging because of room sound, crowd noise, and microphone bleed. Results need to be checked on the actual recording.
Why might instruments remain in the acapella?
Voices and instruments share frequencies and musical patterns. A model can assign a small amount of instrumental sound to the vocal estimate.
Does the demo contain a real singer?
No. The current original demo contains synthesized melody, bass, and drums. It explains stem groups and is not a demonstration of vocal extraction quality.
Can the tool clone a voice?
No. This product is for separating existing uploaded recordings, not cloning voices, changing singers, or generating a new vocal performance.
Keep exploring