Once the bounce is final, video work is no longer “make something that moves.” It is two different output contracts that happen to share an audio file. One file has to stand in for the recording in a feed. The other has to present time-aligned text. If I fold both into a single timeline, I usually get a picture that is neither a clear promotional cut nor a readable lyric cut.
The audio itself stays in Doremi. The decision is which contract to render first.
Contract A: a picture that promotes the recording
YouTube’s partner documentation on music video assets defines a music video asset as a customary music video that embodies one full sound recording of one composition and consists of visual content meant to enhance or promote that recording. It is explicit that this is distinct from an Art Track, the automatically generated YouTube version used for recordings that do not have a produced music video. Art Tracks are also not used for Content ID matching.
I am not delivering a partner package every time I post a demo, but the boundary is still the right design test: does this file exist to enhance or promote the recording, or is it only a placeholder so the player is not empty? If it is the former, I start at the AI Music Video Generator. That page is a type chooser, not a single renderer: Cover Player, Lyric Video, or Music Visualizer, with the cover, the lyrics, or the beat kept in sync. I treat “Lyric Video” on that hub as one available path, not as the default promotional cut.
Contract B: cues with a start time, an end time, and a payload
A lyric video is closer to a caption track than to a storyboard. MDN’s WebVTT API describes Web Video Text Tracks as text cues that are time-aligned with other media such as audio or video. The API is primarily used to display subtitles or captions over video, and each cue has a start time, an end time, and a textual payload. That is the input I need to lock before I care about typography.
The dedicated path is the AI Lyric Video Generator: animated lines that appear in sync with the song, a style, a timing preview, and an export. The page is built for a computer-sized preview; finished files then show up under My Videos on any device.
YouTube’s own help on adding subtitles and captions makes a related point on the platform side: a subtitle or caption file contains a text version of the audio plus timestamps for when each line should appear. A burned-in lyric video is not a Studio caption file, but the same data—the words and their times—is what makes the clip honest. If the times are wrong, the typography is decoration.
I export the two contracts as two files. The promotional cut can change its picture without touching the cue list. The lyric cut can change its style without pretending to be the official video for the recording.
