🤖 Agent Z transcribe podcast

This is an AI Agent to transcribe podcasts to be easily readable as a book 🤘

✨ Features

• Podcast Transcription: Transcribes French tech podcasts into readable text.
• Timestamps: Option to include timestamps in the transcription.
• Download: Save transcription as a Markdown file directly from the browser (compatible with Cloud Run).
• Modern AI: Powered by Gemini 3 Pro.

🛠️ Configuration

Prerequisites:

• Java 25

1️⃣ Export these 2 variables:

export GOOGLE_GENAI_USE_VERTEXAI=FALSE
export GOOGLE_API_KEY=<your gemini API>

2️⃣ Create a .env file. The template .env-template is available.

🧑‍💻 Development

To run your agent (if you have only one specific agent):

mvn dependency:build-classpath -Dmdep.outputFile=cp.txt && java -cp target/classes:$(cat cp.txt) transcribe.ZPodcastTranscribe

Then go to http://localhost:8080/.

🐳 Docker

This app can be run with Docker:

docker build -t agent-z-transcribe-podcast .
docker run -p 8080:8080 -e PORT=8080 agent-z-transcribe-podcast

🚀 Deployment

To deploy on Cloud Run:

1. Authenticate on GCP:

   gcloud auth login

2. Configure settings by exporting variables:

   export GOOGLE_CLOUD_PROJECT=<your project>
   export GOOGLE_CLOUD_LOCATION=<location like europe-west1>
   export GOOGLE_GENAI_USE_VERTEXAI=<true if you deploy on Google Cloud, else false>
   export GOOGLE_API_KEY=<your GOOGLE API KEY>

3. Deploy using gcloud:

   gcloud run deploy agent-z-transcribe-podcast \
   --source . \
   --region $GOOGLE_CLOUD_LOCATION \
   --project $GOOGLE_CLOUD_PROJECT \
   --allow-unauthenticated \
   --memory 1Gi \
   --max-instances 1 \
   --set-env-vars="GOOGLE_CLOUD_PROJECT=$GOOGLE_CLOUD_PROJECT,GOOGLE_CLOUD_LOCATION=$GOOGLE_CLOUD_LOCATION,GOOGLE_GENAI_USE_VERTEXAI=$GOOGLE_GENAI_USE_VERTEXAI,GOOGLE_API_KEY=$GOOGLE_API_KEY"

✨ Contribute

Anyone can contribute to this project. For the moment, please add your question or purpose something in a new issue.

🙏 Contributors

Jean-Phi Baconnais

Benjamin

Guillaume Le Floch

Gudsfile