1
Import a song or supported audio file into Moises and choose the separation or practice workflow you need. Wait for the AI processing to produce stems, then mute or isolate instruments and vocals to focus on the part you are learning. Adjust tempo or key when practicing difficult passages, and export permitted results when your plan supports it. Source separation is not always perfect, especially with dense mixes, so listen for artifacts before using stems in production. Respect copyright and only upload or export material you are allowed to process.
2
For a first evaluation, keep the scope small enough that you can compare the AI-assisted result with a known baseline. Record the configuration, model or workflow choices that produced the result so successful tests can be reproduced. If the product connects to external systems or private data, grant the minimum permissions required and review the provider's current security and data-handling documentation before expanding access.