Mashup Maker — One Song’s Vocals over Another Song’s Music

From the melody, not from key names. The notes the singer actually sings are compared with the chords of the new music at every possible shift, and the shift where both fit one scale wins. On 56 test pairs with known keys it chose the right shift in all 56 when the vocals ran 16 bars or more; with only a few sung bars, the key a fifth away can fit just as well. Up to 2 semitones go to the vocals; beyond that the bass and instruments move too, because a voice shifted further starts to sound processed. The drums are never transposed. Every shift can be chosen by hand, with a bar showing how well it fits.

First the tool checks half and double time: a 70 BPM ballad and a 140 BPM dance track share a pulse, and the vocals then land one beat to two. If the tempos still differ by more than about 12%, the vocals will sound hurried or dragged; “Tempo → In between” splits the change between both songs, and a custom tempo is available too.

It is a verdict given before the build, from what was already found in both songs. “A good pair”: the tempos differ by no more than 6% (half and double time taken into account), and the keys are the same, relative or neighbors on the Camelot wheel — the vocals are barely stretched or transposed and sound the most natural. “A difficult pair”: the tempo has to change by more than 12%, or the songs are in different meters, such as 3/4 and 4/4 — it can still be built, but the voice will sound noticeably processed or slip off the bar. Everything in between “fits with adjusting”: the tool matches the tempo and shifts the key by itself.

Yes. By default the whole vocal is used, from its first phrase, coming in where the music song’s singer came in. Under the timeline, “Fragment of the vocal song” picks where to start — any phrase after a pause — and how many bars to take: to the end, 8, 16 or 32, for example just the chorus. The vocal itself drags along the bars; the arrow keys move it by a bar, Shift + arrow by a beat, and a list suggests other bars where it fits the harmony well.

No. Tick “This file is already an a cappella” above the vocal slot and “This file is already an instrumental” above the music slot — the separation step is skipped, so a mashup can be made even where the separator does not run, for example on an iPhone. An a cappella has no drums, so its tempo is taken from the file name or its tags; if neither has it, the tempo is estimated from the voice and marked “check it”. Turn on the metronome over the a cappella and, if the clicks drift from the singing, fix the tempo with ×2 and ÷2, type the number or tap it. If the finished mashup has the singer slightly off the beat, “Vocal offset” moves the vocal by a fraction of a beat.

Separating a finished song is never perfect: a trace of hi-hats or reverb can stay in the vocals, especially in dense choruses. “Clean up separation leftovers” quiets the vocals between phrases, and the reverb in the mix helps the leftovers blend. Songs with a close tempo and key sound cleanest, because the vocals then need less stretching and transposing.

Yes. Anyone can download the fitted vocals and the music track as separate files — enough to finish the mashup in your own editor. “Mashup stems” are part of Premium: every sounding track — vocals, drums, bass and the other instruments — as a separate WAV file in the mashup’s tempo and key; each starts at the mashup’s first second and runs to its end, so they line up in any DAW. Where the browser can save into a folder (Chrome, Edge and other Chromium browsers on a computer), the files go straight into the folder you pick; elsewhere they arrive as one ZIP archive.

The tool makes the audio; it does not change who owns the songs. Publishing a mashup of someone else’s tracks usually needs permission from the rights holders, and video platforms may mute or block it through automatic content matching. For practice, a DJ set or a personal project it is the same as editing the files yourself.

Yes: “Second vocal — add a third song” opens a third deck. The music comes from song B, the first vocal from A and the second from C, which takes over after the first: when there is enough music, A sings the first half of what is left after it comes in and C the second. Only the voice is separated from the third song; its key is chosen on its own, while the tempo and key of the music are shared. Each vocal moves along the bars and is cut to a section by itself — above the timeline you choose which one you edit; “Remove” takes the second vocal out without rebuilding. Not sure which songs to take? “Find a pair among your songs” compares several at once and shows the best pairs with the same verdict the decks show.

“Video for Reels” under the download button takes a clip of about 30 seconds — whole bars where the vocal sings loudest — and opens it in “Create Video from Audio” with a vertical template and the song titles as the heading. There you pick the background, the visualizer and the text, and the video is saved as MP4. The “What next?” block sends the whole mashup there too.