Bring a reference
Use up to three audio clips to guide a voice or style. Or add an image and describe the sounds of the scene.
AI Audio GeneratorSeed Audio 1.0
Every ideahas a sound.
A voice that tells your story. Music that sets the mood. A scene that feels alive. Create it all from a few words.
Powered by ByteDance · 1,000 audio characters per generation
The listening roomReal Seed Audio samples
Ambience · MP3
Rain after dark
Close-up rain pattering on a window, a low distant thunder roll, then rain easing off. No speech or music.
A taste of its sound effects. Your next creation can include voices and music, too.
One model. So many possibilities.VoiceoversMusicDialogueSoundscapes
Start with an idea
What are we making?
Pick a starting point, change a few words, and take your idea into the studio.
You direct. Seed creates.
A little direction.
A lot of range.
Give the model a feeling to follow, a voice to reference, or a scene to bring to life. Keep shaping until it fits your project.
@Audio1Your reference. Your direction.
Fine-tune the delivery
Adjust speed, volume, and pitch. Choose a target length and a sample rate from 8 to 48 kHz. The actual duration may vary.
Take it into your project
Listen, save, and download. Your creations stay in your history, ready for the next edit, episode, or idea.
MP3WAVOGG OpusPCM
Before you press play
A few useful answers.
What can I create with Seed Audio 1.0?
Create narration, dialogue, music, sound effects, or a scene that combines them. Describe the voices, mood, setting, and how the audio should unfold. Seed Audio 1.0 is ByteDance’s audio creation model.
Can I use a reference voice or image?
Yes. Attach up to three MP3 or WAV audio clips, each up to 30 seconds, or one JPG, PNG, or WebP image. Each file can be up to 10 MB. Use audio references and image guidance separately. Only upload media you have permission to use.
Can I choose the length and download format?
Set a target duration up to 120 seconds, or choose Auto. Duration is prompt guidance, so the actual length may vary. Download MP3, WAV, OGG Opus, or raw PCM. PCM includes a WAV preview for listening.
How much does it cost to generate audio?
Each successful generation uses 1,000 characters from your plan’s audio allowance. You can see your balance and the cost before generating. Listening to the demos on this page does not use your allowance.
How is this different from text-to-speech?
Text-to-speech focuses on reading a script in a selected voice. This generator lets you describe a whole audio idea, including voices, music, ambience, and effects. Use a prompt to direct the performance and scene.