Translate and dub spokesperson videos using Anthropic and deAPI

Who is this for?

- Marketing teams localizing video content for international markets - E-commerce brands creating product videos for multiple regions - Agencies producing multilingual ad campaigns for global clients - Educators and trainers adapting video courses for different language audiences - Anyone who wants to localize a spokesperson video without re-filming

What problem does this solve?

Localizing a video for a new market usually means hiring a local presenter, re-filming the entire video, or settling for subtitles that nobody reads. This workflow takes an existing spokesperson video, transcribes it, translates the speech into a target language, generates dubbed audio, and produces a lip-synced talking-head video with a locally-relevant face — all without a camera or a casting call.

What this workflow does

1. Reads the original spokesperson video and a reference image of the local presenter in parallel 2. Transcribes the video's audio to text using deAPI (Whisper Large V3) 3. Extracts the raw transcript text from the transcription result 4. AI Agent translates the transcript into the target language, preserving tone and pacing 5. Generates dubbed speech in the target language using deAPI text-to-speech (Qwen3 TTS Custom Voice) 6. Generates a lip-synced talking-head video from the dubbed audio using deAPI audio-to-video generation (LTX-2.3 22B), with the local presenter image as the first frame

Setup

Requirements