Nodaro DocsSpeech to Video
Generate a video driven by a speech track with Wan 2.2. Connect speech audio, a portrait and a prompt, and get a talking video from 30 credits.
The Speech to Video node generates a video driven by a speech track. It uses the Wan 2.2 speech-to-video model: connect speech audio, and optionally a portrait and a prompt that describes the scene, and the node returns a talking video that follows the speech.
When to use it
- A video presentation driven by a narration.
- Animated content from a podcast or a voiceover.
- A talking character animated from its voice track.
To make an existing portrait or clip speak, with more model choices, use Lip Sync.
Quick start
Add the node
Press Tab on the canvas and choose Video › Animate & Perform › Speech to Video.
Connect the speech and the portrait
Connect an audio node with clear speech to Audio, for example Text to Speech. Connect a portrait to Portrait.
Describe the scene
Write the Prompt in the settings panel, for example "a presenter speaking in a bright studio", and choose the Resolution.
Run it
Click Run. The video appears on the node.
Inputs
| Input | Accepts | What it does |
|---|---|---|
| Audio | Audio nodes | The speech that drives the video. Required. |
| Portrait | Image nodes | A reference image of the speaker. Optional. |
| Prompt | Text nodes and pickers | The description of the scene. You can also type it in the settings panel. |
| Cinematography | Look and camera pickers, such as Lighting, Lens and Camera Motion | Each picker adds its wording to the prompt. |
The output, Video, is the URL of the generated video.
Settings
| Setting | What it does |
|---|---|
| Resolution | 480p, 580p or 720p. The price of each is shown in the list. |
| Prompt | The scene to show while the speech plays. |
| Negative Prompt | What to leave out of the video. |
| Pre & post text | Text always added before and after the prompt. See Prompt pre and post text. |
Click Show Advanced Settings for fine control. Leave a field empty to use the model's default.
| Setting | What it does |
|---|---|
| Seed (optional) | A fixed number that makes a run repeatable. Empty means random. |
| Num Frames (16-81) | How many frames to generate. |
| FPS (8-24) | Frames per second. |
| Inference Steps (1-50) | More steps give more detail but take longer. |
| Guidance Scale (0-20) | How closely the video follows the prompt. Higher is more faithful, lower is more varied. |
| Shift (0-20) | A model parameter for fine-tuning the result. |
Credits
The price depends on the resolution: 30 credits at 480p, 50 at 580p and 60 at 720p.
Tips
- Use clean speech. A clear, well-recorded voice gives the best result.
- Describe the scene. A prompt next to the audio gives you more control over the picture.
- Preview cheap. Use 480p for quick previews and 720p for the final video.
- Change advanced settings last. Start with the defaults, then raise Inference Steps for more detail if you need it.
Frequently asked questions
Related
Lip Sync
AI Avatar
Text to Speech
Camera Motion
Last updated on
Lip Sync
Make a portrait speak an audio track, or dub an existing video by re-syncing its lips to new speech, with Kling Avatar, OmniHuman, Seedance and more.
Motion Transfer
Make a still character perform the moves of a reference video, such as a dance or a gesture, with Kling 2.6, Kling 3.0 or Wan Animate.