Skip to content
Log inSign up free

AI audio model

Seed Audio AI audio generator

Voices, dialogue, music, and sound effects from one prompt, at exactly the length you ask for.

Try Seed AudioSee the prompt guide

No installation. No credit card. Private by default.

At a glance

Seed Audio is an AI audio model. On Mage it makes audio from a prompt or one reference image in tracks of 5 to 120 seconds. It costs from 3.75 Gems (about $0.004) per second and runs unlimited from the Basic plan, and new accounts get 300 free Gems to start.

  • 3.75Gems per second, from
  • $0.004About, in US dollars
  • BasicUnlimited from
  • 5–120 sLength

Overview

What is Seed Audio?

Seed Audio is the audio model on Mage. It turns one text prompt into speech, dialogue, music, or sound effects, as an MP3 of exactly the requested length.

It can match a voice from audio references you saved, or score a scene from an image you give it.

Samples

Made with Seed Audio

Generated on Mage with Seed Audio. A prompt under a sample is the exact prompt that made it.

  • A narrator line from a single prompt

    Prompt: A warm, unhurried male narrator with a slight rasp says: "Every lighthouse keeps one promise: to stay lit until morning." Small wooden room, faint sea wind outside, no music.

  • A lo-fi music bed with rain and vinyl crackle

    Prompt: Mellow lo-fi hip hop beat with a soft electric piano, vinyl crackle and light rain in the background, 85 BPM, no vocals.

Strengths

What Seed Audio does best

Best at

  • Narration and character dialogue
  • Music beds and loops
  • Sound effects and ambience
  • Audio matched to an image you made

Trade-offs

  • Audio only: pair it with a video model for picture
  • An image reference and audio references cannot be combined
  • Several video models generate their own sound, so a clip may not need a separate track

Prompt guide

How to prompt Seed Audio

Tips that work on this model, then example prompts to start from.

Tips

  • Describe the voice like a casting note: age, warmth, pace, accent, the room it is in.
  • Put the exact words in quotes.
  • For music, name the genre, the instruments, the tempo, and "no vocals" if you want none.
  • For ambience, list the layers: rain, distant traffic, a dog barking once.
  • Pick the duration first: the clip is exactly that long.
  • Mention a saved voice by @handle to keep a character sounding the same.

Example prompts

  • A narrator line

    A warm, unhurried male narrator with a slight rasp says: "Every lighthouse keeps one promise: to stay lit until morning." Small wooden room, faint sea wind outside, no music.

  • A music bed

    Mellow lo-fi hip hop beat with a soft electric piano, vinyl crackle and light rain in the background, 85 BPM, no vocals.

  • Sound effects

    A heavy wooden door creaks open, footsteps on stone, then a torch crackles to life in a cold, echoing hall.

Specs

Seed Audio specs and inputs

OutputAudio
Lengths5, 10, 30, 60, 120 seconds
Image inputsOne reference image
Mentions by @handleAudio references
Where to use itThe Mage app, the Mage assistant, Claude and ChatGPT through the Mage MCP server, and the Mage API

Pricing

Seed Audio price and plans

Every price comes from the same pricing that bills your generations.

ModelSettingGemsPerUSDBasicProPro PlusMax
Seed Audio—3.75second$0.004Unlimited up to 10 sUnlimited up to 30 sUnlimited up to 60 sUnlimited up to 120 s

Prices in Gems, and in US dollars at the list price of the 10,000 Gems pack ($10). Each plan column shows what that plan runs unlimited, up to the longest length it covers; anything else costs Gems. Other settings, such as quality levels and extra input images, can change the price.

Plans and pricing

FAQ

Seed Audio questions

How much does Seed Audio cost?
Seed Audio costs from 3.75 Gems (about $0.004) per second. Gems come with every paid plan and in Gem packs, you pay only for what you generate, and a generation that fails is refunded.
Can I use Seed Audio unlimited?
Yes. Seed Audio runs unlimited from the Basic plan ($10 a month) for tracks up to 10 seconds. Longer tracks can need a higher plan or Gems; the pricing table shows every setting.
Can I try Seed Audio for free?
Yes. New Mage accounts get 300 free Gems, enough for 16 tracks of 5 seconds with Seed Audio. No credit card is needed to start.
What can I give Seed Audio as input?
A text prompt, plus image inputs: one reference image. It takes saved audio references by @handle in the prompt.
What lengths does Seed Audio support?
Lengths: 5, 10, 30, 60, 120 seconds.
Can I use Seed Audio through an API, or in Claude and ChatGPT?
Yes. The Mage API runs it as the seed_audio model, and the Mage MCP server lets Claude, ChatGPT, and other assistants generate with it after quoting the price.
Is what I make with Seed Audio private?
Yes. Everything you generate on Mage is private by default: it appears in Explore or on your profile only if you post it.

Compare

More models on Mage

  • From 20 Gems per image

    Anima

    An anime and illustration model with clean linework, flat color, and tag-based prompting.

  • From 35 Gems per image

    Chroma

    An open-weights, general-purpose image model built on the Flux architecture.

  • From 40 Gems per image

    Flux 2

    The open-weights image model for long, detailed prompts and edits from several input images.

  • From 9 Gems per image

    GPT Image 2.5

    OpenAI's image models: prompts followed to the letter, and text in the image that actually reads.

  • From 75 Gems per image

    Grok Imagine

    xAI's Grok Imagine image model in three quality tiers, from quick ideas to polished finals.

  • From 48 Gems per image

    Guava 2 Pro

    Photorealism that passes for a photograph: portraits, fashion, and product shots with true-to-life skin, fabric, and light.

Make your first audio with Seed Audio

New accounts get 300 free Gems, enough to try Seed Audio. No install, no credit card.

Start creating for free