Can I describe the sound in my own language?

Yes. Write it in English, Chinese, Spanish or most other languages. It is turned into the generator’s words first, and your own sentence is what you see in your library.

How long can a sound effect be?

Each take is 1 to 22 seconds. For longer ambience, turn on Loop and make it longer in your browser: up to 10 minutes as WAV or 60 minutes as MP3, with no seam.

Are the loops really seamless?

Yes. A loop is made so its end runs into its start, the studio plays it round so you can listen for the seam, and it downloads as WAV so the player adds no gap. Made longer, it repeats the same seamless loop.

What file formats can I download?

WAV at 48 or 44.1 kHz in 24-bit stereo, or MP3 at 320 kbps. A loop as made is WAV; made longer, it can be an MP3 too. Every take can also come in one ZIP.

Why do I get four takes?

The same description can sound different each time, and one take from any sound effect generator is rarely the perfect one. Four side by side let you pick the best, or keep them all as variations for a game. You can also ask for 1, 2 or 3.

Where can I find a sound effect I can’t find anywhere?

When no library has the exact sound or length you need, describe it and generate it here: "a heavy iron gate swinging shut in the rain, close up" takes seconds to make.

Can it make the same sound a set number of times?

Counts like "barks twice" or "three knocks" are hard for any generator, and a take may have one more or one fewer. Generate a few takes and keep the one that fits, or cut a take to the part you need.

Can it make music or voices?

The AI sound effect generator makes sound effects and ambience. For music, use the AI music generator; for a voice reading your words, use text to speech.

How do I edit a sound effect after it is made?

In the studio, open Cut and fade on any take: drag its edges, add a fade in or out, and play the cut. For pitch or speed, download it and open it in the pitch changer.

How long are my sound effects kept?

They stay in your library for 30 days, where only you can see them. Download the ones you want to keep for longer.

AI Sound Effect Generator from Text

Q: How much does the AI sound effect generator cost?

One credit for every 3 seconds made: 4 takes of 5 seconds use 7 credits, and 4 loops of 10 seconds use 14. The credits are shown on the button before you start, and a run that doesn’t finish gives them back. Your sign-up credits make about 8 runs of 4 five-second takes. See pricing for credit packs and plans.

side by side from one description: 4 takes
describe it as you think it: Your language
a take; loops run for minutes: 1 to 22 s
seamless loops for ambience: 60 min
silence off, even loudness, no clipping: Ready to use
4 takes of 5 seconds: 7 credits

How do you make a sound effect with AI?

To make a sound effect with AI, describe the sound in a sentence: what makes it, what it is made of, where it happens and how near. Pick a length, turn on Loop for ambience, and the AI sound effect generator returns 4 takes in seconds. Listen, cut the one you like and download it as WAV or MP3.

How to use the AI sound effect generator

Three steps from an idea to a file you can drop on a timeline.

Describe the sound
Write a sentence in your own language, up to 300 characters: "an old wooden door creaks open slowly, then the latch clicks shut". The examples show how a good description reads.
Set the length, loop and takes
Choose 1 to 22 seconds and up to 4 takes. Turn on Loop for rain, wind, a crowd or a hum that should go round without a seam.
Listen, cut and download
Play the takes side by side, cut and fade the one you like, and download a WAV at 48 or 44.1 kHz, an MP3 at 320 kbps, or every take in one ZIP.

What this sound effect generator does for you

A sound effect maker built around the moment you drop the file into your edit: it should sound right, sit at the right level and loop cleanly.

Small waves washing over a pebble beach at dawn

Describe sounds in your own language

Write the description in English, Chinese, Spanish or most other languages. It is turned into the generator’s own words first, so you hear the sound you described.

Takes that are ready to use

Raw AI sound effects come out at very different levels, some with a second of silence at the end and some clipping. Here every one-shot has its silence trimmed and a short fade at the end, loops keep their seam, and every take comes out at the same loudness with no peak over the limit.

Seamless loops, as long as you need

Turn on Loop and the end of the sound runs into its start with no seam you can hear. Then make it longer in your browser: up to 10 minutes as WAV, or 60 minutes as MP3, for a room tone, a storm or a game level.

Four variations, one ZIP

The sound effect generator gives up to 4 takes for each description, so you can choose the best one or keep them all as variations. Download every take in one ZIP, named alike and at the same loudness, ready for a game engine that picks one at random.

Cut and fade each take

Drag the edges of a take to keep just the part you want, add a fade in or out, and play the cut before you download it. The file you get is the sound you heard.

Clear specs and credits

Every take is stereo, as WAV at 48 or 44.1 kHz or MP3 at 320 kbps. You pay one credit for every 3 seconds made, shown on the button before you start, and a run that doesn’t finish gives the credits back.

AI sound effects made from one sentence

Every sample was generated by this sound effect generator from the description on its card and made ready to use, with no other edits. Loops play round, so you can listen for the seam.

Sound effect generator guide: describe a sound so it comes out right

How to describe a sound effect

For AI sound effects that sound right, a good description names the thing that makes the sound, the action, the material and the space around it. "Door" gives you a door; "an old wooden door creaks open slowly in a stone hallway, then the latch clicks shut" gives you the door you imagined.

Say how near it is and how big the room is: "close up", "in the distance", "in a small tiled bathroom", "outdoors in an open field". The space changes the sound as much as the source does.

1What makes the sound: a door, rain, a sword, a crowd.
2What it does: creaks open, drums on a roof, is drawn from a sheath.
3What it is made of: old wood, tin, metal, gravel.
4Where it happens and how near: a stairwell, a forest, close up.

Not sure? Start with the examples in the studio and change one word at a time.

Sound effects for videos and short films

For a video, ask for the sound as the picture shows it: a whoosh as the title slides in, a deep impact on the cut, footsteps on wet gravel as someone walks away. Set the length close to the shot, then cut the take to the frame.

Pick WAV at 48 kHz for video projects: it is the rate most editors and cameras use, so nothing is resampled on the timeline.

Game sound effects and variations

Games play the same action many times, and one sound repeated a hundred times gets tiring. Generate 4 takes of the same description and use them all as variations: the ZIP keeps them at the same loudness with the same name and a number, so a game engine can pick one at random.

Keep game sounds short and tight: a sword swing in 1 or 2 seconds, a coin pickup in 1. Turn on Loop for level ambience, engines and machinery that run for as long as the player stays.

Seamless loops for ambience and backgrounds

Ambience is easier to use as a loop: rain, wind, a café, the hum of a ship. With Loop on, the AI sound effect generator makes the sound so its end runs into its start, and the studio plays it round so you can listen for the seam before you download.

A loop downloads as WAV so it repeats with no gap; make it longer in the browser and it can be an MP3 too, up to 60 minutes, for a stream, a study track or a sleep sound.

Interface and app sounds

Interface sounds should be short, soft and consistent: a tap, a toggle, a success chime, an error. Ask the sound effect generator for "short", "soft" and "clean", choose 1 or 2 seconds, and keep the takes that sit together.

Because every take comes out at the same loudness, a set of interface sounds made one after another already matches. Turn the ready-to-use processing off only when you want the raw take.

When a take isn’t quite right

Change one thing at a time: the material, the distance, the speed. "Slow, heavy footsteps" and "quick light footsteps" are different sounds. Counts are hard for any generator: ask for "a dog barks twice" and you may get three barks, so generate a few takes and cut the one that fits.

If the sound has too much going on, describe less. A single clear event sounds better than a list of five.