duck music
Lower the music under speech: every music clip ramps down just before the first word, holds, and ramps back after the last; pauses under a second stay dipped. With ref (a library item, or a search_stock id imported first), the same call lays it under the whole timeline, its end_at second (default the file’s end) on the last frame, faded out. With level bed (default with ref), speech goes to delivery level first unless within 1 dB or carrying volume keyframes; the music is set by measured loudness 13 dB under it while dipped, below it between lines, peaks under -1 dB. Speech comes from the clips’ words (voiceover and dialogue, not lyrics or music); speech without word timing is analysed from the audio. depth is a fraction of the music’s level: soft to half (default with ref), medium to a quarter (default), deep to a tenth. The response gives each clip’s dips as [start, end] timeline seconds and its volume while dipped, and with ref or level bed the window placed and the music’s loudness against the speech. Dips are volume keyframes; edit_timeline setKeyframes with an empty list removes them. Free, except analysing untimed speech or first measuring a file: cloud processing.

