This project showcases a fully automated pipeline that converts WordPress articles into narrated audio using a combination of AI processing, cloud services, and no-code orchestration.

• The workflow automatically:
• Monitors new posts published on a WordPress site
• Saves the post content to a Notion database for tracking and archival
• Cleans the article text with OpenAI
• Generates a voiceover with Google Cloud Text-to-Speech
• Uploads the audio file to Google Cloud Storage
• Embeds the audio player directly into the original WordPress article

How AI Enhances the Workflow

AI plays two essential technical roles in this system:

1. Text Cleaning with OpenAI
   Before narration, raw HTML content is parsed and cleaned using the OpenAI API. The AI removes tags, links, code blocks, and unwanted symbols while preserving meaningful punctuation and structure. This ensures the voiceover sounds natural and professional.

2. Voice Generation with Google Cloud TTS
   Depending on the size of the article, the voiceover is created using:

• Direct Text-to-Speech (TTS) for short text (≤ 5000 bytes)
• Long-form synthesis via a custom Cloud Run service for longer posts
• This makes the system efficient and cost-effective, handling both short-form and long-form content seamlessly.

Tech Stack Overview

| Component | Description |
| Make.com | Orchestrates the full automation |
| OpenAI API | Cleans raw HTML and formats text for narration |
| Google Cloud TTS | Generates voiceovers (MP3/WAV) |
| Google Cloud Run | Hosts the long-form synthesis API via FastAPI |
| Google Cloud Storage | Stores audio files with public access |
| WordPress API | Fetches and updates blog posts |
| Notion API | Stores articles in a structured database |

How to implement this project:

🧱 Requirements

• Accounts on WordPress, Notion, Make.com, Google Cloud Platform
• WordPress Make.com plugin
• Enabled APIs in GCP:
  • Cloud Storage
  • Text-to-Speech
  • Cloud Run
  • Cloud Build

---

⚙️ Make.com Setup: Scenarios

This automation is divided into two main workflows:

• Scenario 1 – WordPress → Notion
• Scenario 2 – WordPress → Google Cloud TTS (VoiceOver)

---

✏️ Scenario 1/2: WordPress → Notion

📌 Goal

Automatically save each new WordPress post into a Notion database, preserving title, link, author, date, and formatted content.

🪜 Step-by-Step

1. Connect WordPress to Make.com

   • Install the Make plugin on WordPress
   • Copy the API key and paste it into Make.com

2. Connect Notion to Make.com

   • Create an integration in Notion (Internal Integration)
   • Give it access to the database
   • Create a database with these columns: Title, Author, Date, Link, Number

3. Create the Make Scenario

   • Module: WordPress - Watch Posts
     • Type: post
     • Status: all
     • Limit: 150

4. Router

   • Splits the flow between Notion and VoiceOver

5. Module: Notion - Create a Page

   • Adds the post data to Notion DB

6. Module: Notion - Append Page Content

   • Adds cleaned post content to the Notion page

---

🎙️ Scenario 2/2: WordPress → VoiceOver (GCP)

📌 Goal

Take a post, clean the text, convert it to audio using Google Cloud TTS, and update the WordPress post with the audio player.

---

🔁 Full Workflow

1. WordPress – Watch Post
   Monitors newly published articles.

2. Router – Flow Split
   Separates Notion logic from VoiceOver logic.

3. Text Parser – Clean HTML
   Combines:

   {{1.title}} {{1.content}} {{1.date}}
   

4. OpenAI – Advanced Text Cleanup

   • Removes HTML, links, code, special characters
   • Keeps punctuation and structure intact

5. Google Cloud Storage – Upload .txt

   • Bucket: wp-voiceovers
   • Filename: {{1.id}}.txt
   • Content: {{47.result}} (OpenAI output)

6. Custom JS – Check file size (5000 bytes)

   byteLength = input.encode('utf-8').__len__()
   return { "exceeds": byteLength > 5000 }

7. Router – Based on file size

   • ≤ 5000 bytes → Short flow (direct TTS)

   • 5000 bytes → Long flow (Cloud Run API)

---

🔽 Flow A – File ≤ 5000 bytes

8. Google Cloud Text-to-Speech

   • Voice: it-IT-Chirp-HD-F
   • Output: .mp3

9. Google Drive – Upload file

10. HTTP – Get public file link

11. GCP Cloud Storage – Final upload

12. WordPress – Update Post

• Embeds audio player for .mp3:

  <audio controls>
    <source src="https://storage.googleapis.com/wp-voiceovers/{{1.id}}.mp3" type="audio/mpeg">
  </audio>

13. Email – Send notification

---

🔼 Flow B – File > 5000 bytes

8. Text Parser – Extra cleanup

   • Removes \n, newline characters, and invisible symbols

9. HTTP – Call Cloud Run API

   {
     "text": "{{61.text}}",
     "language": "it-IT",
     "voice": "it-IT-Wavenet-B",
     "filename": "{{1.id}}.wav"
   }

10. WordPress – Update Post

    • Embeds audio player for .wav:

      <audio controls>
        <source src="https://storage.googleapis.com/wp-voiceovers/{{1.id}}.wav" type="audio/mpeg">
      </audio>

11. Email – Send notification

---

✅ Result

• Posts are backed up in Notion
• Texts are cleaned with AI
• Audio voiceovers are auto-generated and uploaded to Cloud Storage
• WordPress posts are updated with audio players
• Notification email is sent after each run

---

🧠 Tips

• blueprint.json contains the full exported Make.com scenario
• You can customize voices in GCP Text-to-Speech
• Ideal for blog, news, or voice publishing automation

---

Author: Paolo Ronco
Email: info@paoloronco.it
Project: Make.com + WordPress + GCP TTS VoiceOver + Notion DB