AI Stem Separation
aistemseparation.com
- Free plan
- Web
What AI Stem Separation is
AI Stem Separation is a browser-based service that separates audio into vocals, drums, bass, guitar, piano, and other stems, or into vocal and instrumental tracks. Before processing, users can select 2-stem, 4-stem, or full 6-stem separation. The service says it runs Demucs on Replicate, using htdemucs for 2- and 4-stem jobs and htdemucs_6s for 6-stem jobs. It accepts MP3, WAV, FLAC, M4A, OGG, and WEBM files, as well as YouTube and SoundCloud links; Spotify and TikTok imports are not offered. Users can preview stems, download selected tracks or a ZIP, and export WAV with a minute pack. Charges are based on source audio length, rounded up per job, not on the number of stems downloaded. The free plan includes 5 minutes for use within the month; one-time packs start at $4.99. Pack limits affect file size and duration, and processed jobs are available only for a limited retention period. The service has no VST or AU plugin, but downloaded stems can be imported into Logic Pro, Ableton Live, FL Studio, Cubase, or Reaper.
Who it suits
The service is aimed at musicians, DJs, producers, music teachers, content creators, and podcasters who need separate audio parts. It can also suit users who want to move downloaded stems into a listed DAW.
Good to know, and what to watch
- Choose 2-, 4-, or 6-stem processing
- Accepts six audio file formats
- YouTube and SoundCloud links are supported
- Preview stems and download selected tracks or a ZIP
- Free plan includes five minutes
- Spotify and TikTok imports are unavailable
- No VST or AU plugin
- File and job limits vary by pack
- Processed jobs have a limited retention window
Plans
Every plan and priceFree
Spark
Session
Batch
The verdict
AI Stem Separation offers several stem options and supports common audio files and two link sources. The free allowance is limited to five minutes in the current month, while paid packs are one-time purchases starting at $4.99.
What the maker publishes
- Billing
- Processing is charged by source audio length, rounded up per job, and not by the number of stems downloaded. source
- Data processing
- Uploaded audio and generated stems may be transmitted to cloud hosting, storage, and AI inference providers to provide the service. source
- DAW compatibility
- Downloaded stems can be imported into Logic Pro, Ableton Live, FL Studio, Cubase, or Reaper; the service says it has no VST or AU plugin. source
- Exports and workflow
- Users can preview and audition stems, download selected tracks or a ZIP, and export WAV with a minute pack. source
- Import exclusions
- The service does not offer Spotify or TikTok import. source
- Input formats
- The tool accepts MP3, WAV, FLAC, M4A, OGG, and WEBM files, and supports YouTube or SoundCloud links. source
- Intended users
- The site describes use cases for musicians, DJs, producers, music teachers, content creators, and podcasters. source
- Limits and retention
- File size and duration limits depend on the pack, and processed jobs remain available in My Jobs only within a limited retention window. source
- Models and processing
- The site says it runs Demucs on Replicate, using htdemucs for 2/4-stem jobs and htdemucs_6s for 6-stem jobs. source
- Security and privacy
- The privacy policy says the service uses encryption in transit, access controls, and least-privilege practices where applicable, and does not sell personal information. source
- Separation modes
- Users can choose 2-stem, 4-stem, or full 6-stem separation before processing. source
- Support
- The site lists [email protected] for feedback and support. source
- What it does
- The browser service separates audio into vocals, drums, bass, guitar, piano, and other stems, or into vocal and instrumental tracks. source
Checked against the maker's pages on 8 Oct 2026.