MOSS-SoundEffect-v2.0 — Sound Effect Generation AI Model
Describe a sound in English - a door slamming, footsteps on wood, rain on a window, a magic sparkle, a whoosh - and get a 48 kHz mono WAV of the length you ask for. Each clip is level-normalized to the same peak, so set the balance against narration and music in Audio Mix. To lay several effects on a timeline, generate each one, gather them with Sound Collect and feed the list to Audio Mix's Effects input. Generation takes about the same time for 1 second as for 30, because the model always renders the full 30-second window and trims it.
What it does
MOSS-SoundEffect v2.0 takes one English sentence describing a sound and returns that sound as a 48 kHz mono WAV, from half a second to 30 seconds long.
It is built for effects and ambience - impacts, foley, animals, weather, interface sounds, transitions - rather than music or speech. Every clip comes out at the same peak level, so mixing several of them starts from an even footing.
When to use it
- Foley for a short video: footsteps, a door, keyboard typing, a glass breaking
- Ambience beds: rain, wind in trees, birds in a park, crowd applause
- Interface and transition sounds: a notification ding, a camera shutter, a whoosh
- Cartoon and magic accents: a boing, a sparkle chime
- Several effects placed on one timeline with Sound Collect and Audio Mix
Details
- CategorySound Effect Generation
- Year2026
- LicenseApache-2.0
Compliance & Provenance
- ProviderOpenMOSS Team (open weights) · Specialized
- EU AI Act RiskLimited Risk
- Art. 50 TransparencyRequired — AI-generated outputs are marked
Inputs & Outputs
- DescriptionInputstring
The sound to make, in English - what makes it, what happens, and where. One sound per prompt
- SoundOutputsound
Generated sound effect (48 kHz mono WAV, peak-normalized)
Parameters
- duration_secfloatdefault 3range 0.5–30
Length of the sound in seconds.
Clip length in seconds (0.5-30). Does not change generation time.
- num_inference_stepsintegerdefault 100range 20–200advanced
Diffusion steps. More steps take longer.
- guidance_scalefloatdefault 4range 1–10advanced
How closely to follow the description.
- seedintegerdefault -1advanced
Random seed. -1 draws a new one each run.
Limitations
- Prompts are English. Other languages are not supported by this node.
- Output is mono. For stereo width, place effects in Audio Mix or use a stereo source.
- Speech and music are not what it is for - use a TTS node or Text to Music instead.
- Generated audio is synthetic and has to be marked as such under the EU AI Act
Tags
Alternatives in Audio Models
- Music (Lyria)
Google Lyria: text (and optional image) to music — full songs with vocals and lyrics, or 30-second clips. Needs API key.
- Text to Music (ACE-Step 1.5)
ACE-Step 1.5: generate music from a style prompt, with optional lyrics, key, tempo and time signature. 48 kHz stereo, 5-600s.
- Text to Speech (Gemini)
Google Gemini TTS: turn text into a spoken voiceover (Korean/English). Needs API key.
- Text to Speech (Qwen3 Voice Design)
Qwen3-TTS VoiceDesign: on-device text-to-speech with a voice you describe in words. No API key.
- Text to Speech (Qwen3)
Qwen3-TTS: on-device text-to-speech voiceover (Korean/English). No API key.