AI Podcast Transcription

Podcast Generation and Transcription System for Truth Network

Back End Dashboards
Transcription Dashboard
Podcast Details
Public side showing keywords, AI Based description and transcription tab
Python Script that Rents GPUs from cloud providers, provisions docker image and processes podcasts
1 / 4

Transcription Dashboard

Podcast Details

Public side showing keywords, AI Based description and transcription tab

Python Script that Rents GPUs from cloud providers, provisions docker image and processes podcasts

Architected an internal AI content-pipeline platform to drive continuous SEO-rich content growth. The system monitors ~100 podcast feeds in MySQL, detects new episodes, downloads audio, and performs low-cost GPU inference via API to a GPU cloud provider using a custom Docker image — transcribing via NVIDIA Parakeet-tdt with Groq fallback. From each episode, it generates structured enrichment (1–2 paragraph summaries, SEO-specific keyword clusters, sentiment scoring, and spam classification) using Groq + Llama 3.3 70B, returning normalized JSON objects to downstream systems for indexing and publishing.


Results:

  • Transformed a static site into the destination source for Christian-broadcaster podcast content
  • Grew site from ~7 pages of static content to 150,000+ pages indexed in Google
  • Increased traffic by over 600,000%, with daily visitors regularly exceeding 50,000

Technologies Used

Bootstrap Docker GPU Cloud Provider API Groq jQuery Llama 3.3 70B MySQL NVIDIA Parakeet-tdt PHP Python WaveSurfer.js

Project Details

Completed
September 2025
Status
Active

Links

View Live Project