Prompting Omni Flash with sound: write the audio as its own line
Gemini Omni Flash 1.1 always generates audio. A prompt layout that separates picture from sound, three example prompts, and the request on Sume.

Because Omni Flash always produces native synced audio, the prompt is also your sound brief. A simple layout is: one sentence of picture, one sentence of camera, one sentence that starts with "Sound:". Nothing in the docs requires that format; it is a habit that keeps audio from being left to chance.
Sume's Video Router page states the audio is always on and that generate_audio: false is rejected, and the model takes no audio input, so you cannot supply a voice track to follow.
Three prompts
| Use | Picture and camera | Sound line |
|---|---|---|
| Cafe hook | A barista pours latte art into a white cup. Close, slow push in. | Sound: milk pouring, a spoon on porcelain, quiet cafe murmur. |
| Shoe drop | White sneakers on wet pavement, low angle, static. | Sound: rain on asphalt, one footstep, no music. |
| Tool demo | A cordless drill drives a screw into oak. Side view. | Sound: motor whine rising, then a short stop. |
Request
The sound line goes into the same prompt string.
curl -X POST https://api.sume.com/v1/videos \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: sound-test-001" \
-d '{
"model": "gemini-omni-flash-1.1",
"prompt": "A barista pours latte art into a white cup. Close, slow push in. Sound: milk pouring, a spoon on porcelain, quiet cafe murmur.",
"duration": 5,
"resolution": "360p",
"aspect_ratio": "9:16"
}'Test cheaply
Audio is in the price, so a sound test costs the same as a silent one: $0.1875 for 5 seconds at 360p. Run three wordings at 360p ($0.5625) and keep the one you like before you spend on 1080p.
Sources
Related posts
More in Developers
- Python 3.15 rc3 in CI: test your Sume job poll logic before Oct 9
Python 3.15.0rc3 is out and the final is planned for Oct 9. Add a CI job that runs a stdlib unittest on your Sume status-poll decision function.
- Python argparse CLI that checks seconds per Sume model first
A 25-line argparse CLI: choices reject unknown models, a duration check uses the documented limits, and --dry-run prints the body. Runs offline.
- Python asyncio: one prompt, three Sume video rows, a $1.75 probe
Vidu Q4 is not on Sume. This Python asyncio script sends one 5-second prompt to Wan 3.0, MiniMax H3 Max and Omni with Idempotency-Keys for $1.75 total.
- Python asyncio price check: one Omni Flash clip at four resolutions
A runnable Python script submits a 3-second Omni Flash clip at 360p, 720p, 1080p and 4K together and prints usage.cost. Expect $2.20 in total.
Written by Sume