Music Composition
Write structured musical cues, audition them, and mix them under a scene.
Map musical changes to the edit
HKGAI Music Generation produces an audio asset from tags and song structure. Plan the emotional arc separately from the visual prompt; a music node does not itself assemble a video.
- Mark where the scene opens, changes, and ends; decide whether it needs instrumental underscore or lyrics.
- On empty canvas choose Audio → Music Generation, place the node, and give the cue a recognizable Track name.
Give the piece a beginning and end
The song editor offers [intro], [verse], [chorus], [bridge], [inst], and [outro] insertion chips. Treat the markers as a short arrangement plan, not a guaranteed timeline in exact seconds.
- Enter “[intro] sparse keys; [verse] gentle pulse enters; [outro] resolve and fade;” in Song structure / lyrics.
- Generate once, then audition whether the transition coincides with the intended reveal; revise structure instead of appending contradictory tags.
Treat duration as a result to inspect
The Music Generation panel marks duration as API-controlled and provides an audio preview and download on success. Do not promise a fixed length based only on the desired edit.
- Listen to the complete result for an abrupt ending, unwanted voice, or noise; keep the preferred take.
- Connect its Audio output to Video Edit, enable Background music, and decide whether to loop or let it end naturally.
Check the mix and usage rights
A powerful music bed can mask narration even if both tracks are individually clean. Provider availability and rights to reuse source samples are separate questions.
- In Video Edit compare the original-volume control against the music-volume control, then listen on speakers and headphones.
- Check the license for any imported sample or reference; if Music Generation fails, ask an administrator to verify HKGAI music access without sharing secret values.