ElevenLabs Sound Effects V2 generates sound effects from a text description. On Raywake, the default-options price is up to $0.07 per run. Its inputs include a loop switch, optional duration, prompt influence from 0 to 1 and MP3, PCM, telephony and Opus format choices. Raywake does not list a maximum clip length; test a short clip first.
Use ElevenLabs Sound Effects V2 for a sound you can describe, such as surf hitting rocks, a closing door or a mechanical hum. A clear description of the event is more useful than asking for an unspecified “cinematic” effect. Prices and settings are as of 31 October 2026.
ElevenLabs Sound Effects V2 at a glance
| Setting | Available input |
|---|---|
| Required description | text |
| Duration | Optional duration_seconds, numeric or null |
| Loop request | loop, default false |
| Prompt influence | 0 to 1, default 0.3 |
| Default output | mp3_44100_128 |
| Format families | MP3, PCM, µ-law, A-law, Opus |
| Default-options price | Up to $0.07 per run |
How much does ElevenLabs Sound Effects V2 cost?
This price is the most a run with default options costs. It is not a rate per second of generated audio.
| Request | Listed price | What the amount means |
|---|---|---|
| Default options | Up to $0.07 per run | Catalog upper bound for that configuration |
Do not multiply this amount by a desired clip length. A changed duration or output setting should be quoted as the request you actually plan to use.
Keep a distinction between the sound you want and the configuration you have priced. If you revise a short impact into a sustained ambience, revisit the inputs and quote rather than assuming the original estimate covers the new brief.
How should you describe an AI sound effect?
Start with the source of the sound and what it does. “A wooden door closes with a soft latch click in a quiet hallway” identifies the object, action and acoustic setting. “A dramatic sound” leaves all three undecided.
For an ambience, describe what continues and what changes. A coastal scene might need rolling surf with occasional splashes against rock. A machine might need a steady motor with intermittent metal vibration. These descriptions give you audible details to evaluate.
Say whether an event is close or distant when that matters to the scene. Keep unrelated events out of the first request so you can judge the main effect before adding complexity. These are prompt-writing suggestions, not guaranteed acoustic controls.
When should you enable looping?
Set loop to true when the intended use is a repeating background, such as an environmental bed or a sustained machine sound. It defaults to false, which is a clearer starting point for a single event like an impact or latch click.
A loop request still deserves listening. Put the resulting clip on repeat in your editor and check the join for a noticeable change. The existence of a loop input does not replace review in the context where the sound will play.
The optional duration_seconds field accepts a number or null. You can omit it for an initial request, as in the example below.
How do prompt influence and output format work?
prompt_influence ranges from 0 to 1, with 0.3 as the default. Keep it at the default for an initial request, then change it deliberately if you are comparing how the description is followed.
The default output_format is mp3_44100_128. Other listed choices include pcm_48000 and opus_48000_128, as well as additional MP3, PCM, µ-law, A-law and Opus values. Choose an exact enum value that your receiving audio workflow supports.
Avoid requesting a generic wav string here. It is not one of the listed output-format values, even though another speech or audio model may accept it.
How to call ElevenLabs Sound Effects V2 through the API
The quote example describes a continuous coastal ambience and requests looping. It leaves duration unspecified and uses the listed default format.
curl --fail-with-body https://api.raywake.com/v1/quotes \
-H "Authorization: Bearer $RAYWAKE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"elevenlabs/sound-effects/v2","input":{"text":"Rolling ocean surf with occasional splashes against rocks, no voices","loop":true,"prompt_influence":0.3,"output_format":"mp3_44100_128"}}'The model ID is elevenlabs/sound-effects/v2. The description belongs in text, not prompt. That distinction matters when moving between image, video and audio examples in the same API.
Read the quotes reference for the response flow. This example only requests a quote; it does not generate or download an audio file.
How should you review the finished sound?
Listen once on its own, then under the intended scene. Check whether the main event is recognizable and whether background details compete with dialogue or music. A sound can be convincing alone but too busy in a mix.
For a repeating ambience, audition the loop boundary. For a single event, check its beginning and decay. Make the next prompt describe the specific correction, such as a softer latch or less background rumble.
Common questions
Is this a text-to-speech model?
This model generates sound effects. Use a speech model when you need a script read by a voice.
Can it request a loop?
Yes. Set loop to true; its default is false.
What is the maximum duration?
Raywake does not list a maximum clip length for this model; test a short clip first.
Is $0.07 a fixed charge for every request?
No. A run with default options costs up to $0.07. Quote your actual settings.
Describe and create a sound effect