Add Instrumental to Vocal Tracks

Upload a vocal-only recording, then describe the backing you want with style tags. The accompaniment is generated from the audio you uploaded and the tags you set — drums, bass, chords, and lead parts written under the take rather than around a song you described — and each run returns two versions. There is no free-form prompt box for this workflow; every choice about how the instrumental should sound is expressed as comma-separated tags, which keeps the model focused on serving your existing vocal rather than inventing a new song around it.

One thing is required: the vocal audio file you upload. Add Instrumental has no library-track option — you are supplying the vocal that the instrumental is built around. The other three fields are optional controls in the studio and each one arrives pre-filled, so a run works untouched: positive style tags describing the arrangement, negative tags describing what to avoid, and a title. The file you upload is the whole of what the accompaniment is generated from, so send the cleanest take you have.

20 credits per run, and each run returns two versions. Credits are taken when the job starts and go back to your balance if the run fails. Every engine in the studio picker costs the same.

The backing is generated from the vocal you upload, not from a preset beat, and every run returns two versions, so you can keep the one whose groove sits best with your take. The tightest results come from a dry recording sung at a steady pace, with no reverb or backing baked in. If a version drifts, change one tag and run again for 20 credits.

This endpoint describes the target sound through a comma-separated tags field rather than the longer free-form style description used in Generate Music. Keeping the input as focused tags — for example "lo-fi hip hop, warm rhodes, brushed drums, upright bass" — helps the model apply a single consistent character across the whole backing track. A long prose description can pull the arrangement in several directions at once; tight tags keep the drums, bass, and chords sitting in one genre and one feel.

They are mirror-image features that share one engine picker and one price but solve opposite problems. Add Instrumental builds a band under a vocal you already have; Add Vocals puts an AI singer over an instrumental you already have. Both cost the same, and neither asks you to fill in the tag fields yourself. Use this comparison to pick the right one: Aspect · Add Instrumental · Add Vocals — You upload · A vocal-only take · An instrumental-only track — AI generates · The full backing band · The sung performance — Style input · Tags only (no prompt) · Lyrics, plus optional style and tags — Engines · Same picker, one price · Same picker, one price — Cost · 20 credits · 20 credits

The engine picker in the Add Instrumental studio lists only the engines this job can be sent to — the same list as Add Vocals — and it opens on a default, so you can run without choosing one. Older generations are not withheld as a premium tier; this job is simply not available on them. Every engine in the picker costs the same 20 credits.

The endpoint behind Add Instrumental will not accept a blank negative-tag line, so a value is always sent with your job — but you never have to write one. The studio opens with the box already filled, and a safe default is substituted if you clear it, so a run never fails for an empty field. It is still the field most worth editing, because instrumental backing is easy to over-produce: "distortion, clipping, muddy low end, harsh cymbals" keeps the mix clean, and "lo-fi, tape hiss" pushes toward a modern polished sound. Since you are not writing a positive prompt here, the two tag lines are your only steering controls.

Think in three layers and separate each idea with a comma: genre and mood first ("cinematic, uplifting"), then the rhythm section ("four-on-the-floor kick, punchy snare, sub bass"), then the melodic and textural instruments ("bright piano, layered strings, arpeggiated synth"). Naming the actual instruments you expect to hear gives the AI a much clearer target than a single genre word. If the first result is too busy, remove a couple of instrument tags and regenerate; if it feels thin, add one lead instrument tag rather than several at once.

MP3, WAV, OGG, M4A, FLAC, AAC, up to 200MB and 8 minutes, one file per run. Your own recordings and demos work. Some commercial tracks may be refused. For the cleanest result, upload a dry vocal without heavy reverb or a backing track baked in — the file you hand over is the whole of what the accompaniment is generated from, so it is the one thing worth getting right before you spend a run. There is no song-length setting in this workflow.

Your own recordings and demos work. Some commercial tracks may be refused. An a cappella taken from a released track can be refused, because the check compares your file with existing recordings. If a file matches an existing recording, it is refused and the 20 credits go back to your balance. Tags that name a real artist are refused the same way, so describe the sound instead, for example "breathy indie pop, warm synth pads".

Two. Every run returns two versions of the arranged track to your Music Factory library, for 20 credits in total. It is not instant: the job runs in the background, so you can close the tab, and the two versions appear in your library when they are ready. They differ from each other and from run to run, so listen through before you build the rest of the project on one.

Yes. All Music Factory outputs include commercial usage rights, including a vocal you have backed with Add Instrumental. The rights cover what the tool generates, not someone else's recording. Credits are consumed at the moment you start the job. If you want the backing and the vocal as separate files, open the result in Separate Vocals.

A typical path is to record a vocal, back it here, and keep working on the result in the same library. To swap in your original take, open the result in Separate Vocals — a vocal/instrumental split costs 20 credits — and place your recording over the instrumental in any audio editor. To try a different feel, run the same vocal again with a new tag line; each run returns two more versions. Because every stage stays inside your Music Factory library, you do not export and re-import between tools.

Yes. Each run costs 20 credits and returns two fresh versions, so if the first backing is too busy, too sparse, or the wrong genre, adjust your tags and run it again rather than trying to fix it after the fact. Small, deliberate changes work best: pull one instrument tag out to thin a crowded mix, add a single lead-instrument tag if it sounds empty, or strengthen your negative tags if the low end is muddy. Treating the tags as dials you nudge between runs gets you to the right arrangement faster than rewriting them wholesale each time.