MVSep
Multi-model online AI music separation platform hosting 100+ algorithms including Roformer, Demucs, and ensemble pipelines
Product node overview & trust baseline
Operator provenance, privacy baseline and site-wide pricing belong directly to the product node and are inherited across linked task edgesMVSep (mvsep.com) is an AI music and audio source separation platform built by competitive audio ML researchers. It aggregates over 100 cutting-edge neural models including BS Roformer, MelBand Roformer, Demucs4 HT, SCNet, MDX23C, and proprietary weighted ensembles for isolating vocals, instrumentals, bass, drums, piano, guitar, and strings. Billing is 100% pay-as-you-go credit packs with generous daily free tier: free users get 50 separations per day (MP3 for unregistered, 16-bit WAV/FLAC for registered accounts); one-time credit packs start at $3 (60 credits) providing 32-bit float WAV exports, 100-minute file limits, ensemble models, and full API access without recurring subscriptions.
Open-source Audio ML Community · Launched in 2021
Provenance baseline: 17/20 · Core team consists of international audio demixing competition winners with transparent model documentation and zero misleading claims.
Zero mandatory subscriptions; 50 free separations daily; one-time credit packs from $3 to $1000 (standard models consume 1 credit/min) that never expire.
Pricing transparency baseline: 35/35 · Cancellation:
Separated audio files are retained temporarily on servers for download and auto-deleted; manual deletion supported
Cloudflare
Product-level pricing & rules baseline
Verified plans, billing periods and cancellation terms owned by this product node; task-specific credit costs and free export limits are detailed per capability edge belowSite-wide audio source separation credit packages on MVSep · 2026-11-12
100+ AI models, MP3 output format, 1 concurrent job;Low queue priority, captcha required, MP3 only
100+ models, 16-bit WAV/FLAC exports, no captcha, medium priority, limited API;50 daily limit, 1 concurrent job, no Ensemble models
100-min audio limits, 32-bit float WAV / 24-bit FLAC exports, high priority, Ensemble models, batch mode, full API;One-time credit pack consumed per usage, no subscription
Discounted credit rate with full high-priority processing and ensemble model support
Rules [1]High-volume discount pack supporting high-concurrency jobs and API integration
Rules [1]5 plan / billing-period records, all listed; — means that period is not recorded.
Zero mandatory subscriptions; 50 free separations daily; one-time credit packs from $3 to $1000 (standard models consume 1 credit/min) that never expire.
Cancellation, sources & verification record
Cancellation and refund terms unverified
MVSep Plans & Credit Pricing ↗ · 2026-11-12 Capture 1 ↗
Archived official pricing sources: MVSep Plans & Credit Pricing (2026-11-12) [text snapshot ↗]
Capability map & linked task edges
Whenever a capability passes fixed-fixture testing, an edge is automatically linked between this product node and the task topic for horizontal comparison.
This product node has a verified operator, privacy, and pricing baseline, but currently has no scored task edges with a free native export. See the discovered capabilities or paywall-blocked test records below.
Discovered product capabilities (pending fixture test / edge incubation)
Recorded capability · Not yet scoredThese capabilities have been identified on the product but have not yet undergone fixed-fixture testing. When an emerging capability co-occurs across multiple products, it incubates a new task edge.
Tested task edges (with scores and native artifacts) are strictly separated from discovered untested capabilities (unscored).