# Seed Audio 1.0 > Direct a sound scene with Seed Audio 1.0 on MixVio. Speakers, timing, ambience, and score cues in one text brief—credits shown first, private MP3. - [Canonical Seed Audio 1.0 page](https://mixvio.ai/models/seed-audio-1) - [Official model source](https://seed.bytedance.com/en/seedaudio1_0) ## Route summary - **Provider:** ByteDance - **Media:** audio - **Inputs:** Text - **Output:** Private speech audio - **Native audio:** Yes - **Credits:** 14 credits per take - **Availability:** Available ## Capabilities - **One-prompt sound scenes:** Direct dialogue, ambience, and music or SFX cues in one text brief—then get a single mixed MP3, not a bare TTS read you have to assemble later. - **Timestamp control:** Pin spoken lines to ranges like [2.8s:8.4s] so dialogue lands on the beat you need for picture, ads, or paced scenes. - **Multi-speaker casting:** Name each speaker and turn in one brief so distinct voices, emotion, and pacing land in a single take—no separate track per role on this path. - **Text-defined voices:** Shape delivery with age, accent, tone, and emotion in the prompt—model-selected voice on this route, with no preset picker or reference upload. ## How to use 1. **Write the scene layers:** Follow the formula in order: scene and mood, speakers, emotion, quoted dialogue, then ambience / BGM / SFX. Name the language of spoken lines when it matters. Prefill from an Example or Key Feature if you want—nothing runs until you generate. 2. **Pin casting and timing:** Sharpen each speaker’s age, timbre, and delivery, then pin critical turns with ranges like [2.8s:8.4s] when beats must land on picture or music cues. 3. **Generate, listen by layer, iterate:** Confirm the 14-credit estimate, generate, then listen layer by layer—speech, ambience, and score. Fix the noisiest problem first, change one part of the brief, and re-run before you download. ## Pricing Generation cost varies by the selected input and output controls. 14 credits per take. - [MixVio pricing](https://mixvio.ai/pricing) ## FAQ ### What is Seed Audio 1.0? Seed Audio 1.0 is ByteDance’s audio creation model for sound scenes—not ordinary text-to-speech. Plain TTS reads a script in a chosen voice; Seed Audio lets speakers, timing, atmosphere, and music cues share one brief and return as one mixed take. On MixVio you write that brief, then generate one private 44.1 kHz MP3 with credits shown first. ### What inputs does MixVio support for Seed Audio 1.0? Text only on this published route: one scene brief up to 2,048 characters. There is no voice picker, no audio or image reference upload, and no @AudioN / @Image markers. Describe casting, dialogue, ambience, and cues in the prompt itself. ### What is Seed Audio 1.0 best for? AI video and dubbing, ads and short-form with VO plus SFX or score cues, and podcast or game drafts—anywhere dialogue, space, and pacing belong in one text-led brief. Prefer Eleven v3 when you need named presets and emotion tags on short VO. See Where Seed Audio fits and Best for / Not for on this page. ### How should I write a Seed Audio 1.0 prompt? Follow a director formula: Scene + speakers + emotion + dialogue + ambience / BGM / SFX + timing. Name who speaks and how they sound, put spoken lines in quotes, describe the space and score concretely, and use ranges like [2.8s:8.4s] when beats must land. Vague “make it cinematic” lines underperform—listen, tighten direction, and re-run. ### Can Seed Audio 1.0 do multi-speaker dialogue and timestamps in one take? Yes for directed scene briefs—name each speaker and turn in one prompt, and pin critical beats with ranges like [2.8s:8.4s]. Results vary; review overlaps and pronunciation, then refine. There is no separate multi-track mixer on this published path. ### How many credits does Seed Audio 1.0 use, and what do I get? Each run reserves 14 credits and returns one private 44.1 kHz MP3. MixVio shows the exact estimate before submission. Failed runs cost no credits. ### Is Seed Audio 1.0 free? It is available on the Free plan when your balance covers the displayed quote. Failed runs cost no credits, and Free delivery limits still apply. ### How long can each Seed Audio 1.0 take be? Duration follows the brief—more dialogue and layered cues usually mean a longer file—but MixVio does not publish a fixed maximum length on this route. Plan for one mixed take per generation; split long scripts across takes if you need more room. ### Can I upload a reference voice or image, or choose a preset voice? Not on this published route. Seed Audio 1.0 has no public voice picker and does not accept reference-voice uploads, image references, or cloning. Delivery is model-selected from your text direction—use only scripts and directions you own or are authorized to use. ### Can I export stems or edit an existing audio file? No. Each generation returns one mixed MP3—there is no stem splitter, multi-track export, or upload-to-edit path on this published route. Generate a new take when you need a different mix, cast, or timing. ### How does Seed Audio 1.0 compare with Eleven v3 or music models? Choose Seed Audio 1.0 for directed scenes and timed dialogue from one brief. Prefer Eleven v3 when emotion tags and a published fal voice such as Rachel matter more on short takes. Both models and available music routes can be used when your balance covers the displayed quote. Use the comparison table on this page for live Best for, inputs, output, and credits. ### How does Seed Audio relate to Seedance and Seedream? Seed Audio 1.0 is the audio member of ByteDance’s Seed family—alongside Seedance for video and Seedream for images. Open those model pages from Models when you need matching video or stills for the same brief. ### Is Seed Audio 1.0 good for real-time or live assistants? No. MixVio’s Seed Audio path is asynchronous generation for pre-rendered audio—generate, review, then use the file. It is not a low-latency conversational or live-agent route. ### Can I use Seed Audio 1.0 outputs commercially? Paid-plan outputs may be used commercially, subject to the Terms of Service and current ByteDance terms. You must have rights to every prompt, script, likeness, brand name, and sound direction. MixVio does not guarantee exclusivity or trademark clearance. Review pronunciation and mix before publishing. ### Are my prompts and outputs private on MixVio? Yes. Prompts are routed server-side, and final audio is archived privately to your workspace—not exposed in public URLs or client logs. Keep private prompts and asset IDs out of shared links.