Seed Audio 1.0Dialogue, Sound Design & Music in One Pass
ByteDance's flagship audio model generates film-quality dialogue, sound effects, and music from a single text prompt, and reshapes audio you already have. Reference voices, extending, inpainting, stitching, and editing, all through one serverless API.
One model for generating and editing audio
Full cinematic scenes in a single pass
Describe a scene and Seed Audio 1.0 generates multi-character dialogue, sound effects, music, and atmosphere together, balanced as one finished track. Set genre, environment, mood, and a continuous sound bed, then layer in precise cues. The model adds natural breaths, room tone, and variable cadence so voices feel grounded in a real space.
Consistent characters across every scene
Supply up to three reference clips (30 seconds each) and reuse them across an entire production with @Audio1, @Audio2, and @Audio3. Build a reusable voice library, keep characters sounding identical scene to scene, or clone your own voice for personalized text-to-speech.
Reshape audio you already have
Seed Audio 1.0 is a generator and an editor in one model. Take a finished clip and give it a new ending without re-recording. Both versions here open on the same line — Lily's deadpan "Sir… that's an alpaca" — then the model inpaints two completely different endings in the same voices and room tone. Extend a clip, fill a gap, stitch takes together, or rewrite a single line, turning one source clip into endless variations.
Hear what Seed Audio 1.0 can create
Hit play on any example below, then copy the prompt and try it yourself in the Seed Audio playground. Every clip was generated with a single prompt.
Game-winning goal
"Inside a huge football stadium, with the deafening roar of tens of thousands of fans throughout the background. The commentator (middle-aged male, British accent, rich and penetrating voice, classic sports commentary, extremely exhilarated) shouts in a rapid, soaring, full-throated tone: "OH, HE SCORES!!! WHAT A GOAL! He beats two men and buries it in the top corner — UNBELIEVABLE! The stadium is on its feet!!!" He draws out the word "GOAL" with a voice slightly hoarse from excitement, and the crowd's cheering erupts at the moment of the goal and continues to the end."
Action movie trailer
"[Action movie trailer style. A collapsing underground train tunnel, red emergency lights flashing through dust, tense and urgent.] [A deep rumbling tunnel collapse continues underneath the whole scene, with distant metal groans and loose concrete falling in short bursts.] Mara (female, early 30s, American accent, sharp commanding voice, controlled but urgent, fast pace) shouts, forcing calm over the chaos: "Move now! The tunnel is coming down behind us!" [A steel support beam snaps overhead with a violent metallic crack, showering sparks onto the tracks.]"
Movie trailer narrator
"Create a deep, polished movie trailer narrator voice. In a world where aliens come down from the skies, everything is about to turn upside down for one family living in South Texas."
British radio commercial
"Create a British radio commercial with some intro music and then background music as a female voice actress says "Celebrate with our Summer Getaway trip sale, and right now you can save 100 pounds per person, which is 500 pounds off for a family of five...""
How to build an
audio-first short film
A guided demo on pairing Seed Audio with Seedance 2.0 to build a short film. To go deeper on prompting Seed Audio, read the Seed Audio guide.
Pair Seed Audio with state-of-the-art video
Start with a complete audio scene, then build video on top of it. Explore the Seedance 2.0 and Seedance 2.0 4K models, or jump straight into the text-to-video and reference-to-video playgrounds, for podcasts, short films, trailers, and more, all through one API.
ByteDance's most advanced text-to-video model. Cinematic output with native audio, multi-shot editing, real-world physics, and director-level camera control.
ByteDance's most advanced image-to-video model. Animate still images into cinematic video with synchronized audio, start and end frame control, and motion prompts.
ByteDance's most advanced reference-to-video model. Generate video from up to 9 images, 3 videos, and 3 audio clips with native audio and cinematic camera control.
How to get access to the Seed Audio 1.0 API
One serverless endpoint handles generation and editing. The client manages the submit protocol, streaming status updates and returning the final audio when generation completes.
import { fal } from "@fal-ai/client";
const result = await fal.subscribe("bytedance/seed-audio-1.0", {
input: {
prompt:
"[Action movie trailer style. A collapsing train tunnel, tense and urgent.] " +
"Mara (female, sharp commanding voice, urgent) shouts: \"Move now! The tunnel is coming down behind us!\" " +
"[A steel beam snaps overhead with a violent metallic crack.]",
},
logs: true,
onQueueUpdate: (update) => {
if (update.status === "IN_PROGRESS") {
update.logs.map((log) => log.message).forEach(console.log);
}
},
});
console.log(result.data);
console.log(result.requestId);
Common questions about Seed Audio 1.0
What is Seed Audio 1.0?
Seed Audio 1.0 is ByteDance's flagship audio generation model, built by the ByteDance Seed team. It goes beyond traditional text-to-speech by generating multi-character dialogue, sound effects, and background music in a single pass. It runs in two modes: text-to-speech, with no background sound, and text-to-audio, where you generate cinematic audio alongside the dialogue. It also edits audio you already have.
What is the difference between text-to-audio, text-to-speech, and reference voices?
Text-to-audio (T2A) generates a full audio scene from a text prompt only, with no reference voice. Text-to-speech (TTS) turns written text into spoken words. Text-and-audio-to-audio (TA2A) generates audio from text plus one or more reference voice clips, which you tag in the prompt with @Audio1, @Audio2, and @Audio3. Use T2A for text-only prompts and TA2A when you want characters to match saved voices.
Can I use my own voices or reference voices?
Yes. You can supply up to three reference clips (30 seconds or shorter each) via the audio_urls field and reference them in the prompt with @Audio1, @Audio2, and @Audio3. This lets you build a reusable voice library, keep characters consistent across scenes, or record your own voice for personalized text-to-speech. You are only authorized to use voices for which you have direct permission. See the prompting guide for the full workflow.
Can Seed Audio 1.0 edit existing audio?
Yes. Seed Audio 1.0 is a generator and an editor in one model. It supports audio extending (continuing a clip naturally), inpainting (filling a missing or new section between two clips), stitching (merging separate clips into one seamless track), and editing (rewriting a spoken line while keeping the same voice, tone, pacing, and room sound). This lets you create alternate endings and endless variations from a single source clip.
What languages does Seed Audio 1.0 support?
Seed Audio 1.0 currently supports English and Chinese audio generation, which makes it a strong option for English and Chinese voice workflows today. When writing your prompt, specify the language of the desired output. Broader multilingual support is planned for a July 2026 update.
How long can the generated audio be?
The maximum generated audio length is 2 minutes per generation, and the maximum text prompt is 2,048 characters. With fal's API you can generate all your clips at once and stitch them together, so producing 30+ minutes of audio for a podcast or other media is straightforward. A planned July 2026 update is expected to support up to 10 minutes of audio in a single pass, along with length control.
How do I get started with the API?
Install the fal.ai SDK (Python or JavaScript), grab an API key from your dashboard, and call the bytedance/seed-audio-1.0 endpoint in a few lines of code. The API is serverless, so there is no infrastructure to set up. Check the API documentation for every available parameter.
Can I use Seed Audio 1.0 for commercial projects?
Yes. Content generated through the fal.ai API can be used in commercial projects. Check fal.ai's terms of service for full details on usage rights and licensing.
Get in touch about Seed Audio 1.0
Want to learn more about integrating Seed Audio 1.0 into your workflow? Leave your details and our team will reach out.