Skip to main content
Node ReferenceSFX Generation Node

SFX Generation Node

Generate sound effects with AI.

The SFX Generation node (AI SFX) turns a text description into a sound effect. It is the non-voice counterpart to Audio Generation: instead of narrating speech, it synthesizes a discrete sound such as an impact, ambience, or UI cue. The result is stored on the canvas as durable audio.

SFX is sound, not speech
Use SFX Generation for non-speech sound effects. Use Audio Generation (Text to Speech / Speech to Speech) for narrated voice. To store or replay an existing audio file, use the Audio node.

What it does

  • Resolves a prompt from the node field or a connected text input.
  • Posts a sound-effects request to the shared /api/generate-audio route with mode: "sfx", locked to the current ElevenLabs sound-effects model.
  • Decodes the returned audio, stores it durably, and emits it on the audio-out port.

Provider and model

SFX generation is bound to ElevenLabs. The node type locks the model to eleven_text_to_sound_v2; the prompt and a few sound-shaping options are the controls you adjust.

Controls

ControlFieldNotes
PromptpromptRequired. Describes the sound to generate.
DurationdurationTarget length in seconds, clamped to the 0.5–30 range. The default is 5 seconds when none is set.
LooploopOptional. Produces a seamlessly loopable effect.
Prompt influencepromptInfluenceOptional 0–1 value. Higher values follow the prompt more literally; lower values allow more variation.
Duration is clamped, not free-form
The duration is constrained to the supported sound-effects window. Values below the minimum or above the maximum are corrected to the nearest valid bound rather than rejected.

Outputs

On success the node emits the generated effect on the audio-out port and stores it as durable MP3 media, reusable by downstream nodes and across sessions.

Generate a sound effect

  1. Add the AI SFX node

    Add an AI SFX node. The configuration panel opens automatically with a duration default already set.

  2. Describe the sound

    Type a prompt or connect a text source describing the effect you want.

  3. Set duration and shaping

    Adjust the target duration, enable looping if you need a tileable effect, and tune prompt influence.

  4. Run and reuse

    Run the node to generate the effect. The stored result flows out of audio-out for downstream steps.

Agent and API notes

The SFX executor posts to /api/generate-audio with mode: "sfx". Like the voice modes, the route is rate-limited, canvas-bound for credential scoping, and idempotent per provider operation. The example below is a direct SFX route body.

POST /api/generate-audio — sound effectsjson
{  "mode": "sfx",  "prompt": "heavy wooden door creaking open, echoing hall",  "model": "eleven_text_to_sound_v2",  "duration": 5,  "loop": false,  "promptInfluence": 0.5}

Was this page helpful?