From generation to understanding, all in one place.
Song generation, section-level arrangement, lyric transcription, music encoding, and open tooling share one connected model system.
More Music in You
Bring your lyrics, instrumentation, and song structure. Explore new versions of your music.
Upload any song and MuLa Studio transcribes the vocal melody into editable MIDI notes, detects the chords and aligns every lyric line. Drag notes in the piano roll, swap chords, rewrite lyrics, or describe the change to AI. The model re-sings only that passage from your edited MIDI.
This edit rewrote the vocal melody from 1:00 to 1:38 in MIDI and left the rest untouched.
Switch between Original and Edited while playing. The position stays put so you can compare directly. Drag the roll to seek.
Generation, encoding, transcription, and understanding work together—or plug into a workflow on their own.
Song generation, section-level arrangement, lyric transcription, music encoding, and open tooling share one connected model system.
Start free with daily credits. Paid plans bill monthly or annually, and their credits refresh every month. No download caps on any plan: download your songs as often as you like. API usage is billed separately.
No card needed
Loading…
Loading…
Use the models, inspect the weights and build new creative tools on top. HeartMuLa turns music intelligence into shared infrastructure.
# install the open model library git clone github.com/HeartMuLa/heartlib cd heartlib pip install -e . # lyrics + tags → a complete song song = heartmula.generate( lyrics=lyrics, tags="warm, reflection, pop" )
Refined lyric controllability and music quality, paired with the 2026 HeartCodec release.
Explore the repository →Start with the model family, then make the interface, workflow and sound entirely your own.