Spotify karaoke mode on a Mac

What Spotify does not have

The Spotify desktop app has no karaoke mode. There is no button that lowers the vocal, no instrumental version of a track a click away, and nothing in Settings that changes that. Searches for “Spotify karaoke mode” and “Spotify karaoke desktop” mostly return instructions for other apps, and this is one of those pages. We would rather say so at the top than three paragraphs down.

What Spotify does have is the song you want to sing, already playing. That turns out to be the hard part: the twenty minutes before singing are usually spent finding a backing track, checking it is the right version and downloading it. If the song is playing, that part is done.

Taking the vocal out of what is playing

karanow reads the Mac’s output stream — the sound already on its way to your speakers — rather than talking to Spotify. It runs a separation model on that sound, on your own Mac, and hands you the result minus the lead vocal. You play a song in Spotify as normal, press one button in karanow, and about two seconds later the vocal is gone and the band is still going. On a busy machine that can be closer to three seconds; after that it stays continuous.

The original singer is not thrown away. The same pass produces the vocal as its own track, and a slider blends it back in at any level — useful when you are learning a line, less useful when you are performing it, and you decide where it sits.

Spotify does not need to support any of this, and cannot break it by changing its interface, because karanow never touches the interface. Nothing is imported, nothing is downloaded, and no song has to be in a catalogue, because there is no catalogue.

What Spotify gives karanow on top

Some applications publish what they are playing, and Spotify is one of them. karanow reads the current track and its position, follows along, and can pause, skip forward and skip back from its own window, so you do not have to go back to Spotify to do that mid-song.

Knowing the track also lets karanow draw the notes to aim at. They are read from the original singing on the recording you are playing — not from sheet music or a generic melody — and your own pitch from the microphone is drawn over them, so being flat looks like being flat. The first time through a song the notes are built as the separated vocal goes past; from the next play they are already there, ahead of you. That pitch runway is marked beta in the app, and it is beta here too: it works, and it is not finished.

What you will hear

Real-time separation is not as clean as an offline pass. An offline tool can look at the whole song and take as long as it likes; karanow has a moving 1.5-second window and a hard deadline. On dense mixes, heavy reverb and stacked harmonies you will hear traces of the vocal; on sparser songs the result is close to transparent. For practising it stops mattering within about a verse.

Pausing, dragging the progress bar or skipping to another song throws the old buffer away, and the new position takes about two seconds to come back. That is the price of working live, and it is the only one.

What it is not

karanow produces no file. The backing track exists while it is playing and nowhere else: nothing is exported, nothing is saved, nothing is uploaded. If you want an instrumental to keep, an offline separator is the right tool.

It needs Apple Silicon and macOS 14.2 or later; Intel Macs cannot run it. And it never sends audio anywhere: the sound is processed in memory on your Mac and dropped as the window moves on.