Moises
AI stem separation and practice tool
Visit moises.ai ↗External link. Not endorsed — curated for usefulness.
What is Moises?
Moises is a music production and practice platform that uses AI to separate audio into individual stems, made by Moises Systems, Inc. The tool isolates vocals, drums, instruments, and other elements from any song, allowing musicians to practice, remix, and create with studio-quality results without requiring a physical studio setup.
The platform offers stem separation as its core feature, converting uploaded tracks into isolated vocal and instrumental components in seconds. Beyond isolation, it includes a chord finder to identify harmonic progressions, a speed changer for practicing at different tempos, and lyric transcription that converts vocals to text. The AI Studio feature generates custom stems and backing tracks based on user specifications, while the Voice Studio creates expressive vocal parts modeled from professional singers in various styles. Video recording capabilities allow musicians to capture performances with synchronized audio and video directly within the app, facilitating content creation and sharing.
Moises operates on a freemium model with a limited free tier offering monthly uploads and basic features, while premium subscriptions unlock unlimited uploads and full functionality. The platform is available as a web application, desktop software, and native apps for iPhone, iPad, and Android devices. The service has gained adoption among 70 million musicians globally, including endorsements from Grammy winners like Cory Henry and Eloy Casagrande of Slipknot, as well as endorsement from Charlie Puth, who serves as Chief Music Advisor. Educational institutions including Berklee College of Music incorporate the tool into curricula. Moises has received recognition including iPad App of the Year (2024) and placement as an Apple Design Awards Finalist (2025), along with Microsoft Store Awards recognition. The company commits to training models exclusively on licensed materials and does not use user uploads for model training, addressing musi