A hammer striking an anvil has a distinctive ring. Hot iron meeting water has a hiss. If those sounds are part of the scene you want to make, put them in the prompt.
CNAPS Studio’s Create a Video with Sound card uses LTX-2.5 to generate a five-second video with matching stereo sound, with 1080p output available. Describe the scene, choose its shape, and click Run. The workflow is already prepared.
You get 3 free runs across all CNAPS Studio cards without an account. Sign in to download your result. A free plan is also available after signing up.
One prompt for the picture and the sound
The card’s example describes a blacksmith working in an old stone forge at dusk. The prompt asks for hammer strikes on glowing iron, sparks, and a final plunge into water. It describes the setting too: firelight on stone walls and blue evening light through an open door.
The audio instructions are just as specific. They ask for ringing metal strikes followed by a hiss of steam, ending with “No music, no speech, no narration.” Play the sample below with sound to hear the result.
Picture and sound are generated together, allowing actions such as hammer strikes, splashes, and footsteps to have accompanying audio. You do not need to find a separate hammer effect and place it on a timeline to try this scene. Play the result to check how closely the sound follows the action.
How to create a video with sound
- Open the card. Visit Create a Video with Sound and play the blacksmith example with the sound on.
- Write your scene. Click Clear & write my own. Describe the subject, what happens, and where it happens. Include the sounds you want to hear.
- Choose the video shape. Choose a wide or tall layout, with 3:2 and 2:3 options also available, to suit where you plan to use the clip.
- Click Run. The card lists a typical processing time of 2–5 minutes for a five-second video.
- Watch and listen. Check whether the action follows your description and whether the sound fits. If something is off, revise that part of the prompt and try again.
- Download your video. Sign in to download the result you want to keep.
Keep the action simple and the sound specific
Five seconds is a short scene. One clear action gives you something concrete to review: a cup being set on a table, footsteps crossing a wooden floor, or a wave breaking against rocks.
Instead of writing “add realistic audio,” name the sound. Is it a light ceramic clink or a heavy metal clang? Is the rain tapping against glass or splashing into a puddle?
You can describe the camera as well. A fixed shot keeps attention on the action. A slow push-in can bring the viewer closer. Start with one camera movement rather than several cuts.
For a simple first test, try:
A close-up of a white ceramic coffee cup being set gently on a wooden café table. Steam rises from the coffee in soft morning light. The camera stays still. Audio: a light ceramic clink as the cup touches the table, with quiet café ambience in the background. No music, no speech, no narration.
This is a suggested prompt to try, not a description of the blacksmith sample.
Listen as closely as you look
Play the result with sound before deciding whether to keep it. Does an impact sound occur when the objects meet? Does the background sound suit the setting? Is there any music or speech you did not ask for?
If the scene feels too busy, simplify it. If the wrong sound dominates, make the audio instruction more specific. Keep the parts that worked and change one thing at a time.
Do I need to upload footage or build a workflow?
No. This card starts with text, and the workflow is already prepared. You can try an idea without recording footage or connecting models yourself.
For a closer look at the model behind the card, read our LTX-2.5 guide.
