Your words, or the desk's — the sheet gets locked before anything cooks.
Instrumental
Random visual — auto-make a video
Voice
Bring a voice of your own: upload a clean recording, drop a YouTube link, or record it here. It rides the job as the voice reference — clean recordings make better results.
What do you want to do with it?
New vocal
Cover = your track with a new voice on it. Remix = new beat and new vocals on your track. Got a YouTube link instead? Every tool's Attach row takes one — or paste it right here:
Tools — pick one to work on a track
New vocal
New vocal
Pick a source track above — then tap the waveform where the extension starts.
Pick a source track above — then drag on the waveform to mark the section to replace.
Duration
How many versions
Voices on file in this studio: the house voice and the local Kokoro voice. Add a speaker for two-voice pieces.
No words, no title. Demo Mode builds the beat with the vocal skeleton on it — a demo track you can write real words onto later.
Studio — pick the sound to demo in
Production
Mashups
Voice
Name a sound or a song. Grab digs Soulseek + archive.org and brings back a list — FLAC first, every result labeled with its format, size and source. Nothing downloads until YOU pick one. Picks land in the grab inbox as REFERENCE stock — learn and listen, never product until cleared. For BANK stock (drum kits, breaks, sounds), BANK THE BEST runs it like Flex runs it: the studio hunts on its own, takes the clearest one, banks it, and drops the rest without downloading them.
Recent grabs
The editing desk. Pick a song you made, pull its stems apart and mix them your way — gain, pan, EQ, compression, reverb and stereo width on every stem — then bounce the mix as a WAV.
0:00 / 0:00
Session target BPM
Song A BPM
Song B BPM
Varispeed stretch: small tempo gaps ride clean; pitch-preserving stretch is a later floor.
TimelineDrag across a lane to select a range. Tap a bar cell to toggle that bar.
Lanes are the stems the song actually carries — lead and background vocals ride together in Vocals unless the stem set splits them.
Inpaint — repaint a section
Inpaint: repaint up to 1:00 of a song with new words — double the usual 0:30.
Select a range on the timeline (up to 1:00), then write the new words.
0:00 / 1:00 max
Master
0 dB
100%
The translator desk. Paste a lyric sheet, pick a direction, and the desk works it in four parts — literal, singable, back-translation, and a fit report — so the words change language and the song still sings.
Direction
Counts are working estimates — the desk's fit report is the final word.
The method — four parts, always
Literal
The exact meaning, line by line — lands here.
Singable
The version that still fits the melody — lands here.
Back-translation
The singable version read back in the source language, so any drift shows — lands here.
Fit report
Line-by-line syllable target vs actual, with every bend flagged — lands here.
Singable translation: the words change language, the melody fit doesn't.
Your library
Stems (Vocals / Instrumental) ride each song's ⋯ Options when they exist.
No songs yet — describe one above and hit Create. Two versions per song land here.
Playlists
Video
⏮
▶
⏭🔁
🎵
Nothing loaded
0:00 / 0:00
912 • two versions per song
🚀 Publish — Make Public
One button puts a song on every connected platform at once. Your creations, your publish — the brand block rides every post automatically.
