How It Works
From voice memo to published podcast episode in under 30 minutes — no editing, no mixing, no manual steps.
The Pipeline
Every episode of My Weird Prompts follows the same automated journey. Daniel records a voice prompt, and the pipeline handles everything else.
Record a Voice Prompt
Daniel opens the MWP Recorder app on his phone and talks about whatever is on his mind — a question, an idea, or just something weird he wants to explore. Prompts typically run 1-3 minutes.
The recorder is a lightweight Progressive Web App that captures audio and sends it directly to the pipeline.
Transcribe & Plan
The voice recording is transcribed using AI, then a planning agent analyzes the prompt to determine the episode structure, key topics to cover, and how Corn and Herman should approach the conversation.
The plan ensures the episode covers the topic thoroughly and stays on track with the prompt's intent.
Generate the Script
Gemini writes a full dialogue script featuring Corn, Herman, and any other characters relevant to the episode. The script is structured as natural-sounding conversation with personality and humor.
Scripts typically run 3,000-6,000 words, producing episodes of 15-40 minutes.
Review & Polish
Two automated editing passes refine the script. Pass 1 uses Gemini with Google Search grounding to fact-check claims and ensure depth. Pass 2 polishes flow, removes verbal tics, and optimizes for text-to-speech.
Both passes include shrinkage guards that reject edits if they remove too much content.
Generate Audio
Each line of dialogue is sent to Chatterbox TTS running on GPU compute. Each character has a unique voice clone — Corn's lazy drawl, Herman's eager bray, Daniel's narration, and more.
TTS runs in parallel across multiple GPU workers for speed. The pipeline includes safety checks for segment failures.
Assemble & Publish
Audio segments are stitched together, silence is trimmed, and the final episode is assembled. Cover art is generated, metadata is extracted, and the episode is published to the website, podcast platforms, and social media — all automatically.
A duration gate rejects episodes under 10 minutes to prevent publishing incomplete content.
By the Numbers
Want the Full Technical Details?
The complete pipeline architecture, cost analysis, and lessons learned are documented in the technical white paper.