Whoosh and riser transition sounds from a music prompt on Sume

Sume does not ship a dedicated sound-effects endpoint, so a whoosh or riser has to come from the music router. That works because the router takes a free-text prompt of 1 to 5,000 characters, and you can ask for a build-up that ends in a hit. Google's Gemini API pricing page, read on 2026-10-05, lists Lyria 3.5 at $0.08 per song; Sume's Music Router charges a fixed $0.125 per accepted generation for the same routed engine.

The three steps

• Generate. Ask for a single riser: no melody, rising noise and a short impact on the final beat. The router does not accept duration or duration_seconds, so length is steered by the prompt and you trim afterwards.
• Cut. Use Timeline audio in split mode: up to 20 ranges from one source in a single job, output wav or mp3, flat $0.01 per job.
• Place. Add the cut as a soundtrack or audio spine in a Timeline render, which bills $0.10 per output minute.

A prompt that tends to work

Describe the sound, not a genre: "A two-second rising synth riser with white-noise swell that ends in a single low impact hit, no melody, no drums, no vocals." Expect the engine to return more than you asked for, which is why the split step exists. Generate two or three takes, since each is a separate $0.125 job, and cut the best range from each.

Budget a transition pack of ten sounds: one prompt can yield several usable ranges, so ten ranges from three takes costs 3 x $0.125 plus one split job at $0.01, about $0.385, before any render.

Limits to be honest about

Music models are tuned for songs, so impacts and risers can arrive with a pad or pulse underneath. Cut tight and listen. Lyria output carries a SynthID watermark according to Google's music generation page, read on 2026-10-05. For hyper-specific foley, a library of recorded effects will still be more precise than any prompt.