kling-video-o3/pro/reference-to-video
Pro-tier reference-driven 3-15s video generation from $0.112 per second.
Seed Audio 1.0 is a text-to-audio model from Bytedance that turns written prompts into clear, natural-sounding speech and spoken audio. You can steer the result three ways: a plain text prompt, up to three short reference clips, or a single reference image that informs the delivery.
Output format: Audio only / formats wav, mp3, pcm, ogg_opus / sample rate 8 kHz to 48 kHz.
Seed Audio 1.0 is built for fast, flexible audio synthesis with controllable delivery.
| Parameter | Required | Type | Default | Range / Options | Description |
|---|---|---|---|---|---|
| prompt* | Yes (*) | string | — | Free text | Text to synthesize; cite reference clips in order with @Audio1, @Audio2, @Audio3. |
| voice | No | string | vivi_mixed_en_zh_ja_es_id | Preset voice list | Preset voice used for synthesis. |
| audio_urls | No | array | — | Up to 3 URLs | Reference clips (wav/mp3/pcm/ogg_opus), each up to 30s and 10MB. |
| image_url | No | string | — | jpeg/png/webp, up to 10MB | One reference image; cannot be combined with audio references. |
| output_format | No | string | mp3 | wav, mp3, pcm, ogg_opus | Output audio container. |
| sample_rate | No | integer | 24000 | 8000 - 48000 | Output sample rate in Hz. |
| speed | No | number | 1 | 0.5 - 2.0 | Read speed; 1.0 is normal pacing. |
| volume | No | number | 1 | 0.5 - 2.0 | Output loudness; 1.0 is normal. |
| pitch | No | integer | 0 | -12 - 12 | Pitch shift in semitones. |
Seed Audio 1.0 on RunComfy uses time-based billing that follows the length of the generated audio.
| Billing unit | Rate |
|---|---|
| Per minute of generated audio | $0.075 |
Estimated cost examples
| Duration | Approx. cost |
|---|---|
| 15 s | ~$0.019 |
| 30 s | ~$0.038 |
| 60 s | ~$0.075 |
| 120 s | ~$0.150 |
Follow these steps to get a clean read from Seed Audio 1.0.
Pro-tier reference-driven 3-15s video generation from $0.112 per second.
Create lifelike video motion fast with Seedance Pro for design pros
Produces crisp 1080p AI videos with smart motion logic and speed
Animate a single image into a smooth video with Kling 2.1 Standard.
Use WAN 2.2 LoRA as latest AI tool for realistic video creation from text.
Create lifelike scenes with synced audio and visual fidelity.
Seed Audio 1.0 is a text-to-audio model from Bytedance that turns written prompts into clear, natural-sounding speech and spoken audio. In a text-to-audio workflow on RunComfy, you type what you want voiced and optionally guide the delivery with a preset voice, reference clips, or a single image. It is built for creators who want prompt-driven audio without recording sessions.
Seed Audio 1.0 works well for voiceover, narration, character lines, audio drama snippets, and multilingual reads. The preset voices and delivery controls let you steer pacing, loudness, and pitch to match a scene or brand. Because Seed Audio 1.0 accepts reference audio, it is also useful when you need a consistent target delivery across multiple clips.
With Seed Audio 1.0 you can attach up to three reference clips and cite them in the prompt by order as @Audio1, @Audio2, and @Audio3, which lets you mix or carry a target delivery into the read. Alternatively, you can provide a single reference image to inform the tone, but image and audio references cannot be combined in the same request. Each reference clip should stay short, within about 30 seconds and 10MB, for reliable results.
Content creators, product teams, and localization studios can use Seed Audio 1.0 for explainer voiceovers, app prompts, ad reads, and dubbed audio across languages. Developers can wrap Seed Audio 1.0 into pipelines that generate spoken audio on demand from scripts or campaign copy. The mix of preset voices and reference-driven delivery covers both quick drafts and more specific character work from one interface.
Seed Audio 1.0 takes a required prompt plus optional voice, reference audio, and image inputs, and exposes speed, volume, and pitch controls. Output can be returned as wav, mp3, pcm, or ogg_opus, with sample rates from 8 kHz up to 48 kHz. Exact limits such as clip size and supported formats can vary by provider settings, so check the RunComfy parameter panel before building around them.
Yes. You can prototype Seed Audio 1.0 in the RunComfy AI Playground Web UI by adjusting the prompt, voice, references, and delivery controls until the output matches your target. Once the setup is stable, call the same Seed Audio 1.0 model through the RunComfy API with identical parameters to automate generation from your backend or content pipeline. This keeps iteration in the browser and production runs in code, without changing model behavior.
Seed Audio 1.0 generations consume usd / credits from your RunComfy balance, and based on available provider information the model is billed at $0.075 per minute of generated audio. New users typically get a free trial usd amount to experiment, after which usage follows the Generation rules shown on the model page. For current rates and any mode-specific differences, refer to the Generation section of the Seed Audio 1.0 page on RunComfy.
RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.