This is a GitHub Action workflow to demo AIGC, the pipeline has tested below steps:
- Call OpenAI API to generate a speaking script per a prompt request from json input
- Configure Workload Identity Federation to let Github Actions access GCP via OAUTH temporary token
- Generate speaking audio via Google text-to-speech API
- Call Studio.d-id API to generate a talk (mp3) speaking the above script
- Get the url of the saved mp3 file
- Call Video AI API to let the above person speak out the above script
Refer the details at aigc-pipelie/README.md