Voice model by ByteDance Seed
Seed Speech
ByteDance's natural-sounding text-to-speech with optional voice selection. Write a script and narrate your video.
Craze runs Seed Speech when you ask for it by name.
- Makes
- Text to audio
- Starts at
- $0.032 per 1,000 characters
- Plans
- Every plan, including Free
Seed Speech is ByteDance's text-to-speech model, optimized for natural, conversational narration. On Craze, it reads your script with an optional voice ID.
Why creators choose Seed Speech
Seed Speech prioritizes naturalness over processing steps. Write the script, optionally select a voice, and generate narration.
Zero-shot voice cloning
The Craze route does not accept a reference-audio upload. Use a voice ID if you have one, or generate speech from text without specifying a voice.
Smooth, conversational delivery
The model reads like a person talking, not a system reciting text. Inflection, pacing, and timing flow naturally. That's especially valuable for educational content, interviews, and storytelling where a flat read pulls the viewer out.
Multilingual without friction
Seed Speech handles language switching. Narrate scripts in English and Spanish using the same selected voice ID when available.
Cost-effective for lean projects
Seed Speech's pricing favors creators working on tight budgets or with long scripts. The per-character rate is competitive, making it practical for hourlong audiobooks or documentation.
How to use Seed Speech in Craze
- Write or paste your narration script.
- Optionally select a voice ID; this route does not take a reference-audio upload.
- Name Seed Speech in chat or select it in the Models panel.
- Craze generates the narration, drops it into your project, and you can trim, shift timing, or regenerate if you want a different take.
Seed Speech pricing
Pricing starts at $0.032 per 1,000 characters and varies with script length. Use the calculator on this page to see the total before you generate.
Who Seed Speech is for
Documentary and educational content
Naturalness is everything in docs and teaching videos. Seed Speech delivers clear, engaging narration at a reasonable cost per minute.
Audiobook and podcast production
Creators narrating long-form audio benefit from Seed Speech's conversational delivery and low per-character cost. Narrate a full book in a few generations without breaking the budget.
Multilingual projects and global reach
If your project spans languages, Seed Speech can narrate scripts in each with a selected voice ID.
Character work on a budget
Generate separate lines with different voice IDs for character-driven shorts or game dialogue.
Seed Speech vs Other Models
Qwen Audio 3.0 TTS from Alibaba offers fine-grained emotion control through tags. Seed Speech prioritizes voice selection and conversational delivery for narration.
For expressive English narration, Gemini 3.1 Flash TTS offers directorial prompts. MiniMax Speech 2.8 HD includes non-speech sounds like laughs and sighs. Seed Audio 1.0 accepts audio or image references for speech.
Voice Cloning Limits and Best Practices
Seed Speech accepts an optional voice ID and up to 5,000 characters of text. The text-to-speech route does not accept a recording as a reference input.
For a different voice, choose another available voice ID. Regenerate the same script if you want a different delivery.
Script Length and Generation Limits
Seed Speech handles scripts up to 5,000 characters per generation on Craze. If you need longer content, split it into sections and combine them in your timeline.
For narration in another language, submit text in that language and review the resulting accent. You can select a different available voice ID when needed.
Multilingual Voice Adaptation
Seed Speech supports multilingual text-to-speech. For international campaigns, generate each translated script separately and select an appropriate voice ID when available.
When Seed Speech Fits Best
Seed Speech fits creators who need text narration with optional voice selection. Write the script, choose a voice ID if needed, and generate. For game dialogue, animation, or educational content, generate lines separately and arrange them in the edit.
Seed Speech specs on Craze
| Task | Inputs | Resolution | Aspect ratios | Sound |
|---|---|---|---|---|
| Text to speech | Text prompt | Follows the source | Follows the input | None |
Frequently asked questions
Seed Speech is a text-to-speech model from ByteDance's Seed team. On Craze, it generates narration from text with an optional voice ID.
You pay $0.032 per 1,000 characters. Use the cost calculator on this page to estimate the total before you generate.
every Craze plan, including Free.
The Craze text-to-speech route accepts a voice ID, not an audio sample for cloning in the same generation.
Yes. Seed Speech handles multilingual text input; voice selection uses an optional voice ID.
Seed Speech prioritizes natural-sounding speech with optional voice selection for conversational narration.
Write your script and optionally choose a voice ID. Name Seed Speech in chat or pick it in the Models panel, and Craze generates narration for your project.
