Every music operation, one chat

Any music, taken apart and rebuilt your way.

Attach a song and ask in plain language. Split it, study it, revoice it, rewrite the lyrics. Every result comes back as a file you can take into your own workflow.

Free sign-in. 5 credits to start, then $1 per credit. No subscription.

late-night-demo.wav3:42
Separate the vocals and drums.

Done. I ran two models side by side on the vocal so you can keep the cleaner one.

vocals_model-a.wav3:42
vocals_model-b.wav3:42
drums.wav3:42
Attach audio or describe the job

Three steps. No editor.

  1. 01 — AttachDrop in an audio file, or describe the job.Use the paperclip, or say what you want done. Works the same on your phone.
  2. 02 — ConfirmNothing charged runs without a heads-up.The assistant flags any job that uses credits before it runs. Lyric swaps wait for you to approve a line-by-line diff.
  3. 03 — DownloadGet the finished file as a download link.If a job fails, the credits go back to your balance automatically.
Who it's for

It doesn't replace producing. It shows you the parts and gives you the controls.

Aspiring artists: learn how songs are built

Take a track you love apart and study each piece.

  • stemsHear the vocal, drums and instrumental on their own
  • drumsIsolate the kick, snare, toms, hats and cymbals
  • theorySee the key, tempo, meter, sections and chords with extensions
  • notationGet the melody as MIDI and a chord sheet PDF to play from

Producers: control over every result

Every intermediate stays yours. Nothing runs without a confirmation.

  • seedsStage files and keep every intermediate take
  • compareRun two separation models side by side, keep the cleaner stem
  • lyric swapsRewrite a song for a client, approve the diff before it renders
  • cleanupStrip reverb, split lead from backing, dry vocals in one pass
  • apiCall the same operations from your own code or AI client

Every tool, in the same chat.

Grouped by job. Ask for any of it in plain language.

Separation

01
  • Split a song into stems: vocals, instrumental, drums and more
  • Pick a task preset, or run several models side by side and keep the best result
  • Isolate drum layers: kick, snare, toms, hi-hats, cymbals

Vocals

02
  • Revoice a vocal as a named singer from our roster, no reference audio needed
  • Revoice a vocal to match a reference recording you supply, singing or speech
  • Generate speech in a cloned voice from text you provide
  • One-pass cleanup: remove the instrumental, strip reverb, split lead from backing

Lyrics

03
  • Look up lyrics for released songs
  • Transcribe lyrics from audio, with optional section tags
  • Full-song lyric swaps. You approve a line-by-line diff, then choose: re-sing only the changed lines on the original timeline, or a fresh performance of the whole song

Analysis & notation

04
  • Key, tempo, meter and a section map
  • Melody as MIDI
  • Chord charts with extensions
  • Render a professional chord sheet PDF from any analysis

Audio utilities

05
  • Combine and layer audio files into one mix
  • Improve the sample rate of low-quality audio
  • Stage and verify files before processing
  • Check your credit balance, list recent jobs, mint permanent download links

REST API + MCP server

06
  • The same toolkit, called from your own code or AI client
  • Job lifecycle: create, poll, fetch deliverables
  • Per-use billing against the same credit balance
  • Keys and setup at /developers
Two ways to use it

Here, or inside the AI clients you already use.

Same tools, same credits, one balance.

Standalone assistant

RECOMMENDED FOR MUSIC WORK

Open musicmadepro.com/assistant, attach your audio, done. The full toolkit, no third-party filter. Works on mobile.

Open the assistant

Your AI clients and code

The whole toolkit ships as an MCP server called “Music Assistant” for Claude Desktop, Claude Code and other MCP-capable clients, plus a plain REST API.

Claude DesktopClaude CodeAny MCP clientREST API
Get your keys at /developers →
If Claude or another client refuses a music job

Third-party AI clients run their own content filters and often block operations on music they detect as copyrighted. A lyric swap on a commercial track is the classic false positive. That's their filter, not ours. Our standalone assistant has no such blocker: you decide how you use your files. Bring the job here and it runs.

Example results

Before and after, straight from the assistant.

LYRIC SWAP

“Hey There Delilah” → “Hey There Nadela”

Beforehey-there-delilah.mp30:30
Afterhey-there-nadela.mp30:30

Made as a gift. Only the changed lines were re-sung, on the original timeline.

STEM SEPARATION

Full mix → vocals and drums

Mixfull-mix.wav0:30
Vocalsvocals-only.wav0:30
Drumsdrums-only.wav0:30
VOICE CONVERSION

Your take, a roster singer

Beforeyour-take.wav0:24
Afterconverted.wav0:24

The reference was a 20-second voice memo.

Pricing
$1 = 1 credit

Pay per job. Buy the credits you need, when you need them. That's the whole pricing model.

  • No subscription
  • Credits never expire
  • Failed jobs are refunded automatically
  • 5 free credits for first-time users
  • Payments by Stripe

Your files stay yours.

We never train on your filesYour uploads are used for your job and nothing else.
Auto-delete after 60 daysUploads and results are removed automatically.
Private storageDownload links expire after one hour. Mint a permanent link when you want one.
You approve lyric changesNothing generates until you sign off. Editing is free.

Attach a file. Get the deliverable.