01 / Separate
The voice, alone
HTDemucs, the model behind our vocal remover, lifts the singing out of the drums, bass and the rest.
Free tool
Drop in a song and get its words written out, line by line, with the time each line is sung. The voice is lifted out of the music first, so the words come through clearly. Fix any line, then save plain lyrics, a synced LRC file for music players and karaoke apps, or SRT subtitles for a lyric video. It all runs in this tab, so your song is never uploaded.
Song
Separating the voice gives far clearer words but takes longer, and the first time the page downloads the separation model (181 MB). Untick it for a song that is mostly voice, such as an acapella.
Your song is processed in your browser. Nothing is uploaded.
How it works
01 / Separate
HTDemucs, the model behind our vocal remover, lifts the singing out of the drums, bass and the rest.
02 / Write
Whisper writes out what is sung and when each line starts.
03 / Use
Edit the lines, then save synced lyrics, plain text, or subtitles for a lyric video.
Questions
No. The voice separation and the speech model both run in this tab. The models are downloaded once and kept by your browser.
Singing is harder to write out than speech: stretched words, harmonies and effects can trip the model, and fast rap or heavy distortion can come out wrong. A clear vocal on a well mixed song works best. Read through and fix lines in the box before you save.
A text file of lyrics where each line starts with its time, like [01:23.45]. Many music players, media centres and karaoke apps show the line being sung from it. Put it next to the song with the same file name.
Yes. Paste the lyrics into the karaoke maker, which takes the singer out and shows each line as it is sung.
Lyrics belong to their writers. Writing them out for yourself, to learn a song or sing along, is fine; publishing someone else's lyrics needs their permission. Your own songs are yours to share.