AI Deck

Music AI — A music AI platform built on high-quality stem separation, with lyrics recognition and mastering available through an API

Music AI is a platform that delivers AI models for music and audio as an API. It is operated by the company of the same name, founded in 2019 and also known for Moises, its app for musicians. The company has teams in the United States, Brazil, and Europe, and states that it processes more than two million minutes of audio per day. At its core is high-quality stem separation built on what the company calls ethical AI, with modules for lyrics recognition, chord detection, music identification, metadata analysis, and AI mixing and mastering layered on top. The primary audience is developers who want to embed audio AI features into their own products, and a REST API is offered alongside SDKs for Python and Node.js.

Key Features

  • High-quality stem separation: Extract vocals, bass, drums, guitars, strings, piano, keys, and wind instruments from a mixed track as individual stems. Used for karaoke track generation, remixing, and transcription support
  • Automatic lyrics and chord recognition: Convert sung vocals into text and align them on a timeline. Chord progressions, key, and bass detection are also supported, which helps with practice charts and song analysis
  • Music identification and metadata analysis: An identification module locates tracks in a large catalog, while an analysis module (Cyanite) adds advanced metadata such as genre and mood. Useful for organizing music libraries and improving searchability
  • AI mixing and mastering: A multitrack mixing module (ROEX) is integrated, so the finishing stage can be automated as well
  • Composable workflows: Each module is designed as a building block, and several can be chained into a custom workflow. Template workflows such as music-ai/stems-vocals-accompaniment are also available
  • Developer-facing API and SDKs: A straightforward design in which you create a job by specifying an audio file URL and a workflow name, poll for status, and retrieve the result. Python and Node.js SDKs and an Embedded SDK are provided

Pricing

PlanPriceWhat’s included
Pay as You Go$0 (usage-based)Pay only for what you use. Per-module rates run from $0.0005/min to $1.2/min. 48-hour temporary storage, 2 concurrent jobs
Professional$25/monthIncludes $25 of monthly credit (a limited-time offer). 100GB of permanent storage, 10 concurrent jobs, 5% off all modules
Business / EnterpriseContact salesSLAs, a dedicated integration engineer, volume discounts, and high-throughput processing. Private model creation and bring-your-own-storage are optional

Pricing is current as of August 2026. Check the official site for the latest information.

Pros & Cons

Pros

  • Stem separation quality is high, so you can use production-grade models without training your own
  • Separation, lyrics, chords, identification, metadata, and mastering are all behind one API, removing the need to stitch several vendors together
  • You can start on usage-based pricing, so small-scale evaluation costs little
  • Modules can be chained into workflows, making it easy to assemble processing that fits your use case
  • The company states a rights-conscious training policy, which makes commercial use easier to explain

⚠️ Cons

  • It is fundamentally a developer platform, so it does not suit people who want to finish everything in a GUI (the company’s Moises covers that need)
  • Per-module rates vary widely, so cost can be hard to estimate depending on what you process
  • On the usage-based plan, storage is limited to 48 hours, so retaining results is something you have to design for
  • Concurrency is capped, so large batch processing assumes a higher plan or a direct conversation with the vendor

Comparison with Similar Services

CriteriaMusic AIMoisesAudioShakeLANDR
DeliveryAPI / SDK (for developers)App (for end users)API-centricWeb service
StrengthsSeparation, lyrics, chords, analysis, masteringPractice and performance supportStem separationMastering and distribution support
Typical usersDevelopers embedding audio AI in productsMusicians and learnersMusic industry and rights holdersIndependent artists
OperatorMusic AI (same company as Moises)Music AIAudioShakeLANDR
Usage-based pricingYesMostly subscriptionContact salesMostly subscription

Who Is It For

  • Developers who want to add karaoke generation, part separation, or lyrics display to their own music app or web service
  • Music distribution and stock-audio businesses that want to tag large libraries with metadata to improve search and recommendations
  • Startups that want to avoid the cost of training and operating their own separation models
  • Production and distribution teams that want to automate everything from separation to final polish in a single pipeline
  • Teams that want to try something small first and commit only after seeing results

Summary

Music AI is a developer platform that bundles music-oriented AI features around a core of stem separation. Because lyrics, chords, identification, metadata, and mastering can all be called from the same API, it saves considerable effort compared with wiring together several vendors. On the other hand, anyone who wants to finish the job in a GUI is better served by the company’s Moises, so the right entry point depends on the use case. A realistic path is to validate on usage-based pricing first and consider Professional or above once processing volume grows.

← Blog