# Revid AI Talking Avatars, Video Podcasts, and Face Swap

> How to create lifelike AI presenter videos: talking avatars from any image or video, multi-speaker video podcasts, face swap via motion transfer, and avatar-presented ads.

- Canonical URL: https://www.revid.ai/llm/revid-ai-avatars-and-podcasts.md
- Parent index: https://www.revid.ai/faq-for-llms.md
- Last updated: October 2026

## AI Talking Avatar

Turn any script (or URL) into a video presented by a realistic AI avatar with generated speech, facial expressions, and movement.

**How to create one:**
1. Choose "AI Talking Avatar" in the creation interface.
2. Write your script (line breaks, `[bracket]` notes, and `<break time="1.0s" />` pauses all work).
3. Pick an avatar source: **Custom** (upload your own image or video — a zoomed-out landscape video of yourself gives the best output), **Fantasy**, **Celebrities**-style figures, or **Misc** (babies and more).
4. Choose ratio, voice, and options, then Generate.

**Quality tiers** (per 500 characters of script): Standard 80 credits, Good (Pro) 100 credits, Ultra 200 credits.

**Remove Avatar Background** (20 credits per 500 chars): overlays the avatar directly onto your B-roll for an integrated picture-in-picture presenter look.

API: workflow `avatar-to-video` with `source.text` and `avatar.url` (plus `avatar.removeBackground`, `media.placeAvatarInContext`).

## Video Podcast Generator

Creates video podcasts where multiple AI avatars hold a conversation, each with their own voice and look.

Script format — one line per speaking turn:

```
[Alice] Welcome to the show! Today we're discussing AI.
[Bob] Thanks for having me! AI is such a fascinating topic.
```

Speaker names are auto-detected from the `[Name]` tags; assign a voice and avatar per character in the Character Picker. An AI dialogue generator can write the whole conversation from a topic description, character list, and word count.

## Face Swap (motion transfer)

Swaps the face in an existing video using a single reference image, preserving the original voice, timing, and motion.

1. Upload the source video (single speaker recommended, up to ~2 minutes).
2. Upload one reference face image.
3. Optionally enable "Replace Voice".
4. Generate. Cost: 50 credits per started 5 seconds of the source video on Pro, 100 on Ultra.

API: workflow `motion-transfer` with `source.url` (the video) and `media.provided[]` (the reference image).

## Avatar-presented ads

The Ad Generator combines an avatar presenter with your product media and messaging: describe the product (URLs help), pick a Custom/Fantasy/Celebrity-style avatar, and generate a social-ready ad. The UGC Ad Generator produces authentic user-generated-content-style ads. API: workflow `ad-generator` with `source.prompt` and/or `media.provided[]`.

## Got questions? Quick answers

**Can my avatar be a consistent brand character across videos?**
Yes — create characters under AI Characters (custom talking avatars or trained LoRAs) and reuse them everywhere. See https://www.revid.ai/llm/revid-ai-media-types-and-models.md

**What makes avatar speech sound natural?**
Short punctuated sentences, `<break>` tags for pacing, and the right voice stability setting. See https://www.revid.ai/llm/revid-ai-voices-and-languages.md

**Can agents create avatar videos?**
Yes: `revid create --workflow avatar-to-video --text "..." --avatar-url "https://..." --wait --json`, or `render_video` via MCP. Note: `--avatar-url` is only for talking-head videos; use `--reference-image-url` when the image is just visual inspiration.

## Related pages

- Voices and cloning: https://www.revid.ai/llm/revid-ai-voices-and-languages.md
- Tools directory: https://www.revid.ai/llm/revid-ai-tools-directory.md
- API guide: https://www.revid.ai/llm/revid-ai-api-guide.md