# Generate Sound Effects

Describe a sound, then set tempo, key and loop guidance in the AI Sound Effects Generator: what each control does, what a run returns and costs, and how to get a usable loop.

> Canonical: https://aimusicis.com/docs/voice-and-audio/sound-effects

The [AI Sound Effects Generator](https://aimusicis.com/sound-effects) makes short effects and textures from a description: a door, footsteps, rain, an interface tone or a rhythmic loop. For a song with structure or lyrics, use the [AI Music Generator](https://aimusicis.com/ai-music-generator) instead.

## The controls

| Control       | What it does                                                        | Range                              |
| ------------- | ------------------------------------------------------------------- | ---------------------------------- |
| Prompt        | Describes the sound: its source, action, distance and surroundings. | Up to 500 characters               |
| Tempo (BPM)   | Guides the pace of pulses, percussion and other repeating patterns. | Any (the default), or 40–240 BPM   |
| Musical key   | Gives a pitched effect or musical texture a pitch center.           | Any, or one of 12 keys from C to B |
| Loop guidance | Asks for a sound that repeats, for use as a continuous background.  | On or off                          |

Leave the key on **Any** for sounds with no pitch center, such as rain or a door. Tempo is **Any** until you choose a BPM, which leaves the pace to the model; set it for loops and rhythmic effects, since a single one-shot has little to hold on to.

## Write a prompt that names the sound

Name the source and the action, then where it happens:

- "Heavy wooden door creaking open slowly in an empty stone hallway, with a soft echo"
- "Footsteps on gravel, close to the microphone, walking at a steady pace"
- "Soft rain on a window with warm vinyl crackle, calm and steady"

A mood alone ("something spooky") leaves the model to guess what makes the sound. Distance and surface change the result more than adjectives do.

## What a run returns

One run costs 5 credits. The model chooses each clip's length; the form has no exact-duration setting, so trim a clip in the [Audio Cutter](https://aimusicis.com/audio-cutter) when it must fit a set time. Results appear in the tool's **Results** history, where you can play and download them. A failed run follows the usual credit refund.

## Make a loop that repeats cleanly

Loop guidance asks for a repeating sound but does not guarantee a seamless join. Play the end into the start before you use it. If the join clicks, trim both ends at a quiet point in the [Audio Cutter](https://aimusicis.com/audio-cutter).

The example clips on the generator's **Sample** tab were made with the same model and settings this form sends. **Try this example** loads the prompt, tempo and loop setting that produced each one.

Related: [Credits and Plans](https://aimusicis.com/docs/getting-started/credits-and-plans) · [Audio Editing, MIDI and Exports](https://aimusicis.com/docs/voice-and-audio/audio-editing-midi-and-exports)