Audio Generation

Audio generation

21 audio models for text-to-speech, music, sound effects, voice design, dubbing, and speech-to-speech.

Quick start

gen-ai generate -m eleven-v3 -p "Welcome to Picsart AI Playground."

{ "name": "picsart_generate",
  "arguments": { "model": "eleven-v3", "prompt": "Welcome to Picsart AI Playground." } }

Input types

Type	Meaning	Models
tts	text → speech	Eleven v3, Multilingual v2, Voice Design v3, Gemini 2.5 Flash/Pro TTS, Grok TTS
music	music generation	MiniMax Music v2, Lyria 3 Clip/Pro, Kling T2A
sfx	sound effects	ElevenLabs SFX v2
sts	speech → speech / transform	Eleven STS v2, Multilingual STS, Audio Isolation, Dubbing

Providers

Provider	Models	Highlights
ElevenLabs	10 — v3, Multilingual v2, SFX, STS, Dubbing, Voice Design, Audio Isolation, Voice Previews	The most complete voice suite
Google	Gemini 2.5 Flash/Pro TTS, Lyria 3 Clip/Pro	High-quality TTS + music
Kling	Kling T2A, V2A	Text-to-audio & video-to-audio scoring
MiniMax	MiniMax Music v2	Full original tracks from a prompt
Grok	Grok TTS	Fast natural speech

Common audio parameters

Param	CLI flag	Notes
prompt	-p	Text to speak, or the music/SFX description
voiceId	--voice	TTS voice selection (model-dependent set)
language	--language	Language / locale (model-dependent)

Voice catalogs differ per model — list a model's accepted voiceId values with gen-ai models info <id> --json or picsart_model_params.

Pairs with video

Generate a voiceover or score here, then drop it onto a video you made in the same app — or use a v2a model like Kling V2A to score an existing clip.