Audio Models

Audio Models are the AI models available in the Audio Node's generation panel. The available model depends on the generation mode you select — Text to Speech, Text to Music, Text to Sound Effect, or Voice Design.

Available models by mode

Choosing a model

FishAudio runs locally in Zeemo and is the only Text to Speech model that supports custom voices. Use it when you want to create and reuse a specific voice across your projects — see Adding a custom voice on the Text to Speech page.

ElevenLabs V3 offers high-quality official voices for Text to Speech. It does not support custom voices — for that, use FishAudio.

Music V1 and Lyria 3 Pro both generate music tracks from a text description. Duration ranges from 5s to 300s in 30-second increments, and cost scales with length.

ElevenLabs (sound effects) generates short sound effects from a description, from 0.5s to 30s.

ElevenLabs (Voice Design) generates three voice samples of about 8 seconds each from a description of vocal qualities. Once you find a sample you like, save it as a custom voice for Text to Speech via FishAudio.

Parameters

Tip: The credit cost is shown on the Generate button before you confirm. For Music and Sound Effect modes, longer clips cost more — check the cost before generating long durations.

⚠️

Note: Custom voices created with FishAudio are available across all your canvases, not just the one where they were created.