Webinar highlight reel music: a bed quiet enough for speech

A webinar highlight reel is almost all speech, so the music should be a quiet, even instrumental with no vocals, set low and ducked. On Sume that is one $0.125 generation on the Music Router and a soundtrack block on a Timeline render, which is $0.10 per output minute, so a 2-minute reel is $0.20 to render.

Brief for a speech-friendly bed

Ask for what stays out of the way: "Calm corporate instrumental, soft piano and light pad, steady 90 BPM, no drums, no lead melody, no vocals, even dynamics." Google's music generation page, read on 2026-10-05, lists an instrumental option for Lyria 3.5. Avoid prompts with a build or a drop: a highlight reel has no time for a climax, and a dynamic arc will fight the speakers.

Mix settings

Start with gain_db around -8 and duck_db around 6 to 10, then listen on laptop speakers, since that is where most viewers watch. The docs allow -60 to 12 for gain and 0 to 20 for duck, and the duck follows the audio spine, so the spine must carry the speakers. A reel with several speakers at different volumes may need the same duck_db but separate source levels set with audio.gain_db.

Use fade_out_seconds for a short tail of up to 10 seconds, and run the unbilled plan call to check length before the paid render. If the track is shorter than the reel, set loop on the soundtrack rather than generating a second track.