Appearance
Sound Effects
Sound Effects makes sounds from a description, with up to four takes at once. It can also give a video its sound: footsteps, splashes, doors and crowds timed to what happens on screen.
Make a sound from text
- Open Sound Effects in the sidebar.
- Choose Text to sound.
- In Describe the sound, type what you want to hear, or pick an example.
- Set Length, Prompt influence, Seamless loop and Takes.
- Choose Generate 4 takes (or Generate sound for one take).
Takes are usually ready in a few seconds. Each one has a player and a Download button.
Add sound to a video
- Choose Video to sound.
- Choose Choose a video or drop it here, or drag a video onto it.
- Optionally, Describe the sound. Leave it empty and Cinara works out the sounds from the picture.
- Optionally, describe parts of the video separately (see below).
- Keep Also return the video with the new sound ticked to get the video back with sound.
- Choose Create sound.
It usually takes under a minute for a short clip. You can leave the page and come back.
Describe parts separately
Give different moments their own sound, for example footsteps until 0:04, then a door slam.
- Under Parts of the video, choose Describe parts separately. The video splits into two parts.
- For each part, set when it ends (in seconds) and describe its sound.
- Choose Add a part to split the last part again.
- Choose × to remove a part.
Parts run end to end from 0:00. The last part always ends with the video.
Options (text to sound)
- Length: tick Pick the length for me, or untick it and choose 0.5 to 22 seconds.
- Prompt influence: 0 to 1 (default 0.30). Lower is More creative; higher will Follow the words.
- Seamless loop: makes a sound that repeats without a gap, for ambience or engines.
- Takes: 1 to 4 versions at once.
The result
- Text results show Take 1, Take 2 and so on.
- Video results show the Original video, Sound effects, and Video with sound if you asked for it.
- If some takes couldn't be made, you see "N of M takes couldn't be made, so they weren't charged."
- Edit description (or Another video) takes you back to the form.
- Generate again runs the same request again.
- Recent sounds on the right lists your last 20, marked Ready, Working or Failed.
Limits
- Text descriptions up to 450 characters.
- Sound length 0.5 to 22 seconds.
- 1 to 4 takes.
- Videos from 1 second up to 2 minutes, and up to 2 GB. MP4, MOV or WebM.
- Video description up to 450 characters.
- 2 to 12 parts, each description up to 200 characters.
Credits
Text to sound
- 20 credits per second of sound, per take.
- A set length is charged as set. For example, 5 seconds × 4 takes = 400 credits.
- With Pick the length for me, up to 440 credits per take are reserved (22 seconds), and you're charged for the length made.
- Takes that couldn't be made aren't charged.
Video to sound
- 100 credits per second of video, rounded up. Putting the sound on the video is included.
If a job fails, your credits are refunded.
Tips
- Describe the source, the action and the space: "A heavy wooden door creaking open slowly in an old castle".
- Make 4 takes and keep the best one.
- Raise Prompt influence when results drift from your words.
- Use a set length when the sound must fit a gap exactly.
- For video, describe parts when the sound changes during the clip.
Related
- Video: clips that already come with sound.
- Music: songs and instrumentals.
- Clean Audio: pull a sound out of a recording.
- API Keys: sound effects are also available from the API.