Guides / Library / Seed Audio 1.0
Seed Audio 1.0 on Higgsfield
Audio model from ByteDance. Higgsfield's catalog lists it as: Seed Audio 1.0 — text-to-audio synthesis with optional voice/audio reference or a single image reference.
Specs
| Output | Audio |
| Made by | ByteDance |
| Accepts | reference images, reference audio |
Similar audio models
Eleven v4
Eleven v4 speech generation with up to 10 preset or reference-element voices. Dialogue text is limited to 10,000 characters in total.
Eleven v4 Turbo
Eleven v4 Turbo speech generation with one preset or reference-element voice. All dialogue turns must use the same voice; text is limited to 10,000 characters in total.
Text to Speech V2
Text-to-speech with a selectable engine (ElevenLabs, MiniMax, Seed Speech, Vibe Voice, Cozy Voice) via the variant param; preset or reference-element voices (voice_type + voice_id).
Qwen Audio 3.0 TTS Flash
Qwen Audio 3.0 TTS Flash with expressive instructions and a selectable system preset or cloned reference-element voice.
Guides that use this kind of model
Sources
Last checked 18 November 2026. Higgsfield changes models, prices and menus often, so confirm on the live site.
Up to 50% off a Higgsfield plan
Higgsfield gives people who arrive through an affiliate link up to 50% off. The offer runs for three hours from the moment you log in, so pick your plan before you click.
Affiliate link. I earn a commission if you subscribe, at no extra cost to you. Terms checked 18 November 2026.
