Lacuna generates complete songs – vocals, instrumentation and production – from a text description, supplied lyrics or an audio input, in under 30 seconds. Around the generator sit lyric writing with a rhyme generator and syllable counter, stem separation for any uploaded track, an online MIDI editor for note-level work, remix, extend and regenerate for specific sections, and automatic BPM, key and tempo detection.
The MIDI editor is what separates this from the one-shot song generators. Generated music is almost right and then wrong in one specific bar, and without note-level editing the only remedy is regenerating the whole thing and hoping – which loses the parts that worked. Being able to fix eight notes rather than roll the dice again is the difference between a novelty and a tool.
Section-level regeneration follows the same logic and is equally welcome. Free credits start with no card, then subscription or credit purchase, with credits that never expire and full copyright ownership for paying users – though no figures appear on the landing page. Generated vocals remain the most obvious tell in AI music, copyright ownership granted by a vendor does not settle what the model was trained on, and separating a commercial track gives you stems without giving you any right to release them.
