Clone a voice and create talking avatar videos using Anthropic and deAPI

Who is this for?

• Content creators who want a consistent on-screen avatar without filming themselves
• Marketing teams producing personalized video messages at scale
• Educators building video lessons with a virtual presenter
• Anyone who wants to turn text into a talking avatar video using a cloned voice

What problem does this solve?

Creating a talking-head video normally requires a camera, lighting, and a person on screen. Voice cloning adds another layer of complexity. This workflow handles everything — provide a short voice sample and an image, type what you want the avatar to say, and get a lip-synced talking avatar video.

What this workflow does

1. Reads a short reference audio clip and a first frame image in parallel
2. Clones the voice from the reference audio using deAPI (Qwen3 TTS VoiceClone) and generates new speech from the provided text
3. Merges the cloned audio and the first frame image into a single item
4. AI Agent crafts a talking-avatar-optimized video prompt — focusing on lip sync, facial expressions, and natural movement — then boosts it with the deAPI Video Prompt Booster tool using the first frame image for visual context
5. Generates a talking avatar video synced to the cloned speech using deAPI (LTX-2.3 22B), with the image as the opening frame and the AI-crafted prompt guiding the scene

Setup

Requirements

• n8n instance (self-hosted or n8n Cloud)
• deAPI account for voice cloning, prompt boosting, and video generation
• Anthropic account for the AI Agent
• A short reference audio file (3-10 seconds, MP3/WAV/FLAC/OGG/M4A)
• A first frame image for the avatar (PNG/JPG)

Installing the deAPI Node

• n8n Cloud: Go to Settings → Community Nodes and toggle the “Verified Community Nodes” option
• Self-hosted: Go to Settings → Community Nodes and install n8n-nodes-deapi

Configuration

1. Add your deAPI credentials (API key + webhook secret)
2. Add your Anthropic credentials (API key)
3. Update the File Path in the "Read Reference Audio" node to point to your voice sample
4. Update the File Path in the "Read First Frame Image" node to point to your avatar image
5. Edit the Set Fields node with your desired text, video prompt, and language
6. Ensure your n8n instance is on HTTPS

How to customize this workflow

• Change the AI model: Swap Anthropic for OpenAI, Google Gemini, or any other LLM provider
• Adjust the avatar style: Modify the AI Agent system message to target different visual styles (cartoon, realistic, professional, casual)
• Add audio transcription: Insert a deAPI Transcribe Audio node before voice cloning and pass the transcript as refText for improved cloning accuracy
• Change the aspect ratio: Switch from landscape to portrait for mobile-first content or square for social media
• Add a last frame image: Use the optional lastFrame parameter in Generate From Audio to control how the video ends
• Change the trigger: Replace the Manual Trigger with a Form Trigger, webhook, or Airtable trigger for batch avatar generation
• Add delivery: Append a Gmail, Slack, or Google Drive node to automatically deliver the generated video