skills.rest
inference-sh
inference-shOFFICIAL
Explore all skills by this author →
WebsiteGitHub

ai-podcast

Generates multi-speaker talking head podcast videos via portrait, TTS, avatar, and merge pipeline.

723|101|Updated Jan 31, 2026
Search Tags:#text-to-speech#podcast#video generation#inference.sh#avatar animation#character creation#talking head video
One-click install
npx skills add https://github.com/inference-sh/skills --skill ai-podcast
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ai-podcast
Source: https://github.com/inference-sh/skills/tree/main/guides/content/ai-podcast
Command: npx skills add https://github.com/inference-sh/skills --skill ai-podcast

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Creating a multi-person talking head podcast video normally requires cameras, actors, voice talent, and editing software. This Skill orchestrates the full AI pipeline — character image generation, text-to-speech, avatar animation, and video stitching — so you can produce a complete multi-speaker video from a script alone.

Core Features & Use Cases

  • Character Creation: Generate real humans (with identity-consistent Phota profiles), brand mascots, or illustrated characters, including alternate angles and logo placement on clothing.
  • Voice & Script Production: Write natural conversational scripts with duration targets, then generate TTS audio with per-character voices and controlled speaking rates.
  • Avatar Video & Merge: Animate each character frame with its audio into talking head clips, then stitch all clips into a final MP4 video.
  • Use Case: A marketing team wants a 60-second two-host podcast-style promo video. The Skill generates both host portraits, produces approved TTS voices, animates each turn sequentially, and merges the clips into one finished video.

Quick Start

Ask the agent to create a 60-second two-person podcast video about your product, starting by generating the two host characters and proposing voice samples for approval.

Frequently Asked Questions about ai-podcast

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create an AI talking head podcast video?▼

Generate character portraits first, write a conversational script, produce TTS audio per turn with inworld/text-to-speech-2, animate each frame with pruna/p-video-avatar, then merge all clips with infsh/media-merger into a final MP4.

How do I keep a real person's face consistent across video shots?▼

Train a Phota identity profile with phota/train using 30-50 face images, then reference the profile_id in phota/generate for every new angle. Gemini image generation does not preserve human facial identity and will produce a different person.

Can I run avatar video generation clips in parallel?▼

No. pruna/p-video-avatar clips must run sequentially because parallel runs compete for GPU memory and fail with CUDA OOM errors. Each clip takes 15-90 seconds depending on audio length.

What speaking rate should I use for podcast TTS voices?▼

Use 1.05 as the default speaking rate for natural podcast pacing and 1.1 for short reactions. Never go below 1.0, since slower rates sound artificial and disengaging. Keep the rate consistent per character.

Why does my podcast video show empty chairs or extra microphones?▼

Wide shots reveal empty seats when only one speaker is on screen, so use tight chest-up framing for individual turns. Always QA frames for extra people, duplicate mics, and wrong logos before the expensive video generation step.

Related Skills

Explore More Content & Communication Skills→
DatTran26DatTran26

ckm:design

Generate brand identity, design tokens, UI styling, CIP mockups, and HTML presentations.

Community
Advanced
sonyho2715sonyho2715

copywriter

Generate AIDA or PAS marketing copy with clear CTAs for digital campaigns.

Community
Intermediate
yudameyudame
2

podcast-cover-art

Generate branded podcast cover art with Gemini via OpenRouter.

Official
Intermediate
loveoplay2023-hueloveoplay2023-hue

camsnap

Capture still frames and video clips from RTSP and ONVIF cameras.

Community
Intermediate