Paste a YouTube link or upload an MP3: the AI takes the lead vocal out, keeps the band and lights the words up as they are sung. Record yourself over it, take after take, and save your performance as a lyrics video.
Pro allows 15 min on Standard · 6 min on High Quality
A karaoke track is a song without its lead singer, with the words on screen at the moment they are sung. The karaoke maker builds both from a finished recording: a neural network separates the voice from the band, and a speech model reads the words off the separated voice and times each one to the music. You do not need a karaoke version of the song, a lyrics file or any software, only the song itself.
The result opens in a karaoke studio in your browser. The lyrics light up word by word, you can sing and record over the band, and every take can become a track of its own. When you are done, download the instrumental, your mix, or a video with the lyrics lighting up in time.
20 seconds of a royalty-free song, split by the Ultra engine on this page. Switch tracks while it plays: the words light up as they are sung. Every track plays at the same loudness.
Song: “Eternal Light” by mahbubjoarder (Pixabay Content License), separated on this site; only the loudness is matched.
Read live from the current plan settings, so this section always matches what the tool enforces today.
See the plans and current prices
The words are not looked up in a database: they are transcribed from the separated vocal of the song you uploaded, so they match this recording, including the live versions and the remixes. Each word is then aligned to the music, which is why the karaoke screen can light it up as it is sung rather than a whole line at a time.
Word-by-word timing works in 16 languages; songs in any other language still get their lyrics, timed line by line. If a word comes out wrong, edit the line or its timing in the studio, and go back to the original at any time.
Recording is built into the studio. Press Record and a three-beat count-in starts the music, then your voice is captured on its own track, clean, over the separated band. Every take can become a track of its own (Take 1, Take 2 and so on) that you can rename, mix, mute and download. You can also upload a take you recorded elsewhere.
Each track has volume, pan, mute, solo and an effect rack with one-tap presets such as Polish, Warm, Hall and Slapback, so your voice can sit in the mix like the original. Add your own video too, and the download becomes a video of your performance with the mix under it.
Studio recordings of pop, rock, R&B and hip-hop separate very cleanly. Very reverberant recordings, live bootlegs and songs where a synth doubles the voice can leave a faint trace of the singer in the instrumental.
Yes. Choose the High Quality engine and its Karaoke model: it removes only the lead singer and leaves the backing vocals and harmonies in the music, the way a professional karaoke version sounds. It is a direct download, without the studio and the lyrics screen.
Word-by-word timing in 16 languages: Arabic, Chinese, Dutch, English, Finnish, French, German, Greek, Hungarian, Italian, Japanese, Persian, Polish, Portuguese, Russian and Spanish. Songs in any other language still get their lyrics, timed line by line.
Yes. From the studio, download the mix with synced lyrics as an MP4: the words light up in time over the music, in the screen style you picked. Download everything instead to get every track as well.
The separation runs on Mazmazika's servers, so a phone browser is enough; the karaoke screen, recording and the mixer all work with touch. For a party, open the studio on a laptop connected to the TV and switch the karaoke screen to full screen.
Singing at home, practice and private study are fine. The original recording keeps its copyright, so performing in public, streaming or publishing your version needs the rights holder's permission.
