Session log
Everything the app did: playback attempts, lyric fits, renders, errors. If something misbehaves, download the log and attach it in chat. It holds the full fit and render diagnosis, line by line.
Drop Suno stems here
Name files by instrument: drums.wav, bass.wav, guitar.wav, keys.wav, synth.wav, pad.wav, brass.wav, strings.wav.
Everything runs on this device. Audio never uploads.
Export
The ZIP holds a Type 1 multi track MIDI at 960 PPQ, a groove_template.mid, song.groove.json with velocity and timing metadata, and FEEL.txt. Open the .mid in Logic with File, then Open, so Logic adopts the tempo and key from the control track.
Default is 0%. Your exact timing, the groove, stays untouched.
How to fix notes
Click a note in any piano roll to hear it and select it. Drag left or right to move its timing. Drag up or down to change pitch, or on the drum roll to change the drum piece. Press Delete, or use the delete button on the stem, to remove a selected note. Solo and mute work per stem for both the MIDI preview and the original audio.
Vocal Recreator
The performance is the driver. This tab captures the melody, timing, phrasing, and dynamics of a Suno vocal stem, lets you mount your own lyrics onto that performance, and renders a new dry vocal performance as a source for Kits.ai conversion.
Dry render. No effects. No reverb, no delay, no chorus, and none of the original Suno audio is blended into the output. The render carries the original dynamics, so no blending is needed. Judge it after Kits conversion, not as a finished vocal.
Vocal stems are captured in full, up to five minutes per performance.
No vocal stems yet. Add a vocal stem, or retag a stem as vocal, in the Stems to MIDI tab.
Dynamics envelope, sampled every 15 ms from the source performance. Swells, falls, and breaths are part of the performance and are rendered, not flattened.
Lyrics
Paste the full lyric sheet here, then parse. Sheets pasted straight from Suno work: section tags like [Verse] and [Chorus] anchor the fit, bracketed delivery notes like [talkbox vocal] and standalone production notes in parentheses are kept as annotations, not sung, and short parenthesized ad-libs like (Yuh!) still count as sung. Every row has a Make note / Make sung button, so if a line is classified wrong, flip it and that section refits. Section tags are optional. Supported tags: [Intro], [Verse], [Pre-Chorus], [Chorus], [Post-Chorus], [Hook], [Refrain], [Bridge], [Interlude], [Outro], [Instrumental]. Tags are never sung. They only help keep each section anchored to its own part of the performance, and a repeated tag with no lines under it reuses the first block with that name. Without any tags, plain lines map in order, line by line.
Render
This render is a performance source for Kits.ai conversion, judged after conversion, not a finished vocal.
Voice print
Voice print: Built-in, Phillip's studio print 2026-10-12Built-in defaults from Phillip's studio print (2026-10-12) are loaded. Upload a recording or build a print in the Voice Print tab to override them.
Optional quick print: upload one clean, bone dry recording of your voice, 30 to 60 seconds is ideal, any length works. This is only the optional voice print, not your song stems. Or build the full fingerprint in the Voice Print tab. A print built there takes over automatically, ahead of the built-in defaults.
Full print protocol: about 5 to 10 minutes, bone dry WAV, no processing or compression on the way in. Sustained vowels ah ee eh oh oo at low, mid, and high pitches, a few seconds each. A slow sung scale or glissando across your range. One verse and chorus sung at real performance effort. About 30 seconds of natural speaking.
Voice Print Studio
Nine short recordings build your full voice fingerprint: range, formants, vibrato, breath, effort, and the runs you actually sing. Record right here with the mic, or record in Logic and drop the file into each section. Everything is measured on this device. Nothing uploads anywhere. The only file that ever leaves is the small voice_print.json you can download at the bottom, a few KB of numbers.
Sections complete: 0 of 9.
Takes and measurements are saved on this device, so you can do a few sections now and finish later.
Your voice print
No sections measured yet. Partial prints are allowed. Missing sections are listed as missing, never filled in with guesses.