Can I upload audio or use songs already in my Library?

Yes. Sign in and upload an audio file, choose a Library resource or type @ to search your songs. Attach the actual recording when requesting an edit or conversion so the Agent can identify the intended source. Library search and metadata inspection are free; paid processing steps require an approved plan. Only use recordings you have permission to process.

Does the AI music agent spend credits automatically?

Paid workflows require your approval of the proposed plan and credit quote before execution. Review source tracks, settings and any maximum estimates first. Your conversation is saved, so refreshing the page does not erase the transcript. Reopen it to check processing status and completed outputs before approving more work.

Music Agent

Create music with an AI music agent

Describe a song, write lyrics, or work with your audio. Review the plan and credit cost before paid steps run.

Music request

Music Agent uses AI to assist creation. Review generated audio in your workspace.

ABOUT THIS TOOL

AI music agent for songs, lyrics, stems and MIDI

Turn a music idea into a plan you can review in chat. MUSICTOOL AI Music Agent connects song generation, lyric writing, stem separation and audio-to-MIDI conversion with the audio in your Library. Describe the result you want, attach a reference when needed, and review the proposed steps and credit quote before approving a paid workflow.

Output: Completed workflows can produce songs, lyric drafts, separated audio stems or downloadable MIDI files. The conversation keeps references to its source audio and results in your Workspace.

How it works

01Sign in, choose your Workspace and describe your music idea. Include the genre, mood, instruments and whether you want vocals or an instrumental. For work on an existing track, upload your audio, choose a Library resource or type @ to find a song.
02Review the proposed workflow, source material, parameters and credit quote. You can ask for lyrics, a new song, an extension, a passage replacement, stem separation or MIDI conversion. Check that the plan matches your request before approving paid actions.
03Follow progress in the conversation and listen to completed song results. Open lyric drafts or download MIDI when those steps finish. Return to the saved chat to reference a result, adjust future steps or continue developing the same project.

Key features

Create music through conversation: describe a song or instrumental, write lyrics, or request changes to eligible tracks already in your Library.
Connect audio workflows: prepare a plan that generates music, separates stems and converts a selected audio part to MIDI in dependency order.
Use your own references: search Library songs, attach uploaded resources and inspect available metadata without inventing missing BPM or key values.
Review before spending credits: inspect the plan and quote, approve paid work, and return to saved conversations and completed results.

Settings explained

Workspace: A Workspace groups the conversation with its music project. Choose it before starting so new results and source references stay in the right project. Reopening a saved conversation restores its associated Workspace rather than starting a separate chat.
Audio resource: Resources are audio references you own in MUSICTOOL. Upload a recording or choose existing Library audio to give the Agent a specific source. Use @ song search when referring to a Library song. Naming a track in plain text alone may not identify the intended recording; select the actual source before approving an edit.
Credit approval: The quote shows the credit budget for the proposed paid steps. Review it together with the parameters before approving. Some audio-dependent steps show a maximum quote because the duration of an upstream result is not yet known. Changing future steps can require a revised quote and approval.

From a music prompt to a reviewed creative workflow

A useful music prompt describes the intended result in terms you would give a collaborator. For a new instrumental, name the style, mood and instruments, and say where the music will be used. For a vocal song, explain the subject and language of the lyrics. You can start with a lyric draft and then ask the Agent to use it in a song-generation plan. For an existing recording, attach the source and describe the part you want to change. Requests such as extending a song or replacing a passage depend on the source track and the capabilities available for it. The Agent prepares the workflow, but you review the actual sources and settings before paid execution. A multi-step request can connect a generated song to stem separation and then a selected stem to audio-to-MIDI conversion. Later steps use the completed output of their dependencies; they cannot produce a result before the required source is available. This is useful when you want to develop a musical idea across several tools while keeping the instructions and results together. Use the standalone music generator when you prefer to choose generation settings directly, or open the individual audio tools for a focused conversion. The chat is another way to prepare and follow those operations, with the same need to listen to and review the resulting music.

Work with Library audio and continue saved music projects

Your Library provides the concrete recordings behind a request. An uploaded reference, a generated song and a separated stem can have different uses, so choose the source that matches the operation you want. Ask the Agent to find a song or inspect a resource before preparing a workflow if you are unsure which item to use. Metadata inspection reports stored information; an absent BPM, key or scale remains unknown. It does not replace audio analysis or guarantee that an inferred musical description is correct. After a workflow starts, its status and completed outputs appear in the conversation. A song result can be played there, a lyric draft can be opened for editing, and a successful MIDI conversion provides a download. Audio-to-MIDI produces note data for further work in music software; it does not preserve a recording as an identical sounding audio file. Listen to separated audio and inspect converted notes before using them in a production. Saved conversations let you return to a project without retyping the full brief. Reopening a chat restores its transcript and reconnects to its runtime, while previously completed outputs remain available as references. If your direction changes, review adjustments to steps that have not started. Stopping a plan does not undo completed work or necessarily cancel an operation already sent for processing. A failed step may need a retry or different parameters; check the displayed status before submitting another request.

Common questions

You can request song generation, lyrics, song extensions, passage replacements, covers, mashups, vocal removal, stem separation and audio-to-MIDI conversion. Available operations depend on the selected source and its eligibility. Describe your goal in chat, then review the proposed plan. Completed outputs can be used as references for later requests in the same conversation.