Clumi AI

Clumi AI is an audio editing tool. Browser based AI vocal remover that separates vocals from instrumentals in songs and videos, with free processing up to 10 minutes per file and a Pro tier offering 30+ stem separation, 60 minute files and priority processing.

Clumi AI is a browser based vocal remover built around one job: pulling a voice out of a finished mix and leaving the backing track intact. The free tool accepts MP3, WAV, MP4 and MOV files, runs the separation automatically, and returns an isolated vocal track alongside an instrumental. Nothing is installed and no account is needed for the basic workflow, which keeps the entry cost at zero for anyone who only needs an occasional acapella or backing track. Output quality is the part the site leans on hardest. Separation produces WAV files at 24-bit depth, aiming for natural sounding instrumentals rather than the smeared, watery results older phase cancellation tricks produced. Processing is described as completing within minutes for most files, and the interface previews the original, the vocals and the instrumental side by side before any download. A single click exports both tracks, and the download format is selectable, with CSV export available for track lists. The free tier is deliberately capped. Files are limited to 10 minutes and 100 MB, which covers most singles but rules out long mixes, DJ sets, live recordings and full podcasts. Pro Separation lifts those caps to 60 minutes and 1000 MB and expands beyond the two stem split into a 30+ instrument model that can peel out drums, drum kit, bass, guitar, piano, lead versus rhythm guitar, acoustic versus electric guitar and other parts, with priority processing on top. Supporting tools cover lead and backing vocal splitting, karaoke track creation, echo and reverb removal, noise cleanup for voice recordings, and a speech versus music versus sound effects split aimed at video editors cleaning up dialogue. Clumi AI suits DJs building edits and mashups, producers hunting reference stems, singing teachers and students working on pitch and timing, language learners who want the voice alone, and video creators who need dialogue without a music bed. Alternatives like vocalremover.org and x-minus stop at two to four stems and generally skip video input; Clumi's own comparison table puts lalal.ai closer on stem count, with Clumi ahead on file duration and size at the Pro level.