Lesson introductions
Prepare a short talking-head segment to introduce a lesson. Keep the speech focused and review it beside the learning material.
Astorie canvas workflow
AI avatar video generator
Bring a portrait and speech audio to a compatible avatar model on the canvas. Choose the model, inspect its inputs and review the resulting clip. The illustration shows the workflow; it is not a playback demo.
Build an avatar video on the canvas
- 1
Prepare a permitted portrait
Use a clear single-person portrait with an unobstructed face. Obtain the necessary permission before using a recognizable person.
- 2
Prepare the speech audio
Use an appropriate speech recording or a separate speech-generation workflow. Review the pronunciation, pace and audio quality before using it to drive the portrait.
- 3
Supply the model inputs
Choose a compatible talking-avatar model and provide the portrait and audio using its current input controls. Check its limits and credit requirement before running.
- 4
Review the complete take
Watch and listen to the result. Check mouth motion, facial details, timing and unwanted movement; revise the input or request another take where needed.
Plan the presenter and the delivery together
Product explainers
Use a permitted presenter portrait and approved speech to explore an introduction. Review any product or brand claims before distribution.
Recurring hosts
Reuse an approved portrait as a reference for later requests. Compare every new take; the same input does not guarantee identical appearance or motion.
What this avatar workflow needs
| Input | A portrait and speech audio, subject to the chosen avatar model requirements. | |
|---|---|---|
| Output | A generated talking-head video requiring visual and audio review. | |
| Where it runs | The native Astorie canvas; this page does not replace the model input controls. | |
| Model differences | Supported formats, duration, language handling and costs vary. Read the active model details before generating. | |
| Identity and voice | Use material you have permission to use. Voice cloning is not a required step in this workflow. |
Talking-avatar questions
Is this the same as generating an avatar portrait?
No. A portrait is a still image. A talking-avatar request also uses speech audio in a compatible canvas workflow to generate video.
Can I use my own voice recording?
Use a recording you have permission to use, in a format accepted by the chosen model. Check the current input requirements and review the recording before generation.
Will the lips and face be exact?
No exact-match guarantee applies. Inspect lip sync, facial details, expression and timing throughout the generated clip.
Which length or resolution will I get?
Those limits depend on the selected model and settings. Check the current model details instead of assuming one fixed output for every avatar request.
Plan a talking portrait
Bring a clear portrait and reviewed speech audio to the canvas.