Seed Audio 1.0 AI Generator
Generated on PixelDojo with Seed Audio 1.0. Produced by PixelDojo's generation pipeline.
Seed Audio 1.0 bills 0.1 credits per second of audio, but the charge is set before the job runs, from an estimate of your prompt's word count, not from the actual clip that comes back. There is also a 10-second floor, so the cheapest possible job is 1 credit even if the prompt is a single word. This page reads the exact formula out of the tool's own code so the pricing is not a guess.
How the billing and inputs actually work
The estimate formula
Word count divided by 2.5 words per second, plus any CJK characters divided by 4 characters per second, rounded up, with a hard floor of 10 seconds. That estimate, not the real output length, is what gets charged.
Why the floor matters
A short prompt like 'Generate a suspense radio drama' is an instruction, not a script, and can produce a much longer scene than its own word count implies. The 10-second floor and the estimate-based billing mean you are not charged extra for that gap after the fact.
The absolute minimum charge
0.1 credits per second times a 10-second floor works out to 1 credit as the practical minimum for any real request.
Three input modes, one endpoint
Speak from the prompt alone, clone a voice from up to 3 reference audio clips addressed as @Audio1 through @Audio3, or guide the audio from a single reference image. Audio references and an image reference cannot be combined in the same request.
Format and delivery controls
Output as mp3, wav, pcm or ogg_opus, at a sample rate from 8000 to 48000 Hz, with speed and volume each adjustable from 0.5x to 2x and pitch shiftable plus or minus 12 semitones.
There is no voice preset list
A legacy voice field is still accepted for compatibility but does nothing. Describe the voice you want in the prompt itself, or clone one from a reference clip.
0.1 credits a second, a 10-second minimum, and the charge is locked in before the audio is generated.
Why Choose Pixel Dojo for Seed Audio 1.0
Professional-quality results with cutting-edge AI technology
0.1 credits per estimated second
The rate is transparent and fixed; only the estimated duration moves the price.
Voice cloning is built in
Up to 3 short reference clips, no separate cloning product or extra charge.
It generates scenes, not just lines
A single instruction can produce a multi-voice, multi-sound-effect scene rather than one flat narration track.
How It Works
How a Seed Audio job gets priced before it runs:
Write the prompt as speech or as an instruction
A script gets read back closely; an instruction like 'generate a short radio drama' gets interpreted more freely and can produce more audio than its own word count suggests.
The platform estimates duration first
Word count over 2.5 words per second, plus CJK characters over 4 per second, rounded up and floored at 10 seconds, sets the credit charge before generation starts.
Run it in the app or over the API
From the tool page, or POST to /api/v1/models/seed-audio/run with your key. Add reference clips for a cloned voice or a reference image to steer the scene.
Loved by creators on PixelDojo
Real feedback from people using PixelDojo, pulled from our in-product surveys.
Top notch quality and strong prompt adherence.
A well resourced sunscription with attention to updates.
Easy to use and there's lots of options.
super lora download
Great variety of tools to create and modify images and video.
Lots of Tools, and the one time I needed support I got help right away
Explore more AI tools on PixelDojo
AI Tools
Compare & Switch
- Best AI Image Generators
- Best AI Video Generators
- Midjourney Alternatives
- Civitai Alternatives
- Runway Alternatives
- Leonardo Alternatives
- Pika Alternatives
- Luma Alternatives
- Magnific Alternatives
- Veo Alternatives
- Flux Alternatives
- Freepik Alternatives
- Seedance Alternatives
- Seedream Alternatives
- Pixverse Alternatives
- GPT Image Alternatives
- Synthesia Alternatives
- Playground Alternatives
- NightCafe Alternatives
- Canva AI Alternatives
- ElevenLabs Alternatives
- ComfyUI Alternatives
- Fal Alternatives
- Replicate Alternatives
Common Questions
Everything you need to know about Seed Audio 1.0
What does Seed Audio 1.0 cost?
0.1 credits per second, estimated from your prompt's word count before the job runs. There is a 10-second minimum, so the smallest realistic charge is 1 credit.
Is the price based on the real audio length or an estimate?
An estimate, calculated up front from word count and any CJK character count. The provider does not report the real output duration back to the platform, so the estimate is what gets charged and it stands even if the actual clip runs longer.
Can Seed Audio clone a voice?
Yes. Send up to 3 reference audio clips, each up to 30 seconds and 10MB, and address them in the prompt as @Audio1, @Audio2 or @Audio3.
How is this different from a dedicated text to speech tool?
Text to Speech is built for reading a script cleanly in a chosen voice. Seed Audio 1.0 handles that too, but also produces full audio scenes such as radio dramas or podcasts, clones a voice from a reference clip, and can take a reference image to steer the sound. It is the broader, instruction-driven tool of the two.
How do I call Seed Audio from the API?
POST to /api/v1/models/seed-audio/run with your API key and a prompt. Add audio_urls for voice cloning or image_url for image-guided audio, but not both in the same request.