☰

The barrier to entry for content creation has never been lower. With the advent of sophisticated Large Language Models (LLMs) and high-fidelity text-to-speech tools, it is now possible to launch a professional-sounding podcast without a microphone, a studio, or even a human voice.

This guide outlines a three-phase workflow to create a high-quality “faceless” AI podcast channel on YouTube using free-to-use tools.

Phase 1: Developing the Script with NotebookLM

The foundation of any good podcast is the script. While many AI tools can generate text, Google’s NotebookLM is uniquely suited for this because it allows you to “ground” the AI in specific source material, ensuring your podcast is informative and accurate.

1. Feed the Brain

Start by creating a new notebook in NotebookLM. Upload the sources you want the podcast to cover. This can include:

  • PDFs of research papers or articles.
  • URLs to specific websites.
  • Copied text from news reports or personal notes.

2. Use a Custom Script Prompt

While NotebookLM has a built-in “Audio Overview” feature, it generates a finished audio file that you cannot edit or control. To create a customized YouTube show, you need a text script you can manipulate.

Instead of clicking the Audio Overview button, go to the chat interface and enter a specific prompt to generate a dialogue.

The Recommended Prompt:

“Based on these sources, write a lively 10-minute podcast script between two hosts, Alex and Sarah. Make it conversational, include interruptions, laughter cues like [laughs], and use simple language. Focus on [Insert Your Key Topic Here].”

3. Review and Export

Once the script is generated, review it for flow. Ensure the “chemistry” between Alex and Sarah feels natural. When satisfied, copy the text into a document.

Phase 2: Generating Realistic Audio with ElevenLabs

To turn your text into audio that listeners will actually enjoy, you need high-quality synthetic voices. ElevenLabs currently offers some of the most human-like AI voices available.

1. Use the “Projects” Feature

Do not use the standard text-to-speech box on the ElevenLabs homepage. Instead, navigate to the Projects tab. This feature is designed for long-form content and allows you to manage multiple speakers within a single timeline.

2. Assign Voices and Contrast

Paste your script into the Project editor. You will then assign specific voices to your hosts:

  • Host A (Alex): Assign a stable, authoritative male voice (e.g., “Adam”).
  • Host B (Sarah): Assign a contrasting, energetic female voice (e.g., “Mimi”).

Having distinct vocal profiles makes it easier for the audience to follow the conversation.

3. Fine-Tune the Performance

AI voices can sometimes speak too quickly. You can adjust the pacing by adding extra spaces or using the “stability” and “clarity” sliders in the settings. If the script calls for a pause, simply add a line break. Once the settings are dialed in, render the audio and download the final MP3 or WAV file.

Phase 3: Creating the Video and Visuals

Since YouTube is a video platform, your audio needs a visual component. For a faceless podcast, the goal is to create a professional aesthetic that keeps the viewer engaged.

1. Generate Host Personas

Use an AI image generator (such as Nano banana pro or similar tools) to create consistent faces for Alex and Sarah. You want these to look like professional headshots or high-quality avatars that represent the “vibe” of your show.

2. Design the Layout in Canva

Open Canva and create a “YouTube Video” project.

  • Place your generated host images on opposite sides of the screen.
  • Design a background that looks like a modern podcast studio or a clean, minimalist office.
  • Add text overlays, such as the podcast name or the episode title.

3. Assembly in ClipChamp

ClipChamp (a free video editor often built into Windows) is the best tool for the final assembly:

  • Import Media: Upload your Canva background and your ElevenLabs audio.
  • Audio Visualizer: This is a crucial step. Add an “Audio Visualizer” overlay from the graphics menu. This creates a moving waveform that reacts to the voices, signaling to the viewer that they are watching an active conversation rather than a static image.
  • Sync and Export: Ensure the visualizer matches the length of the audio. Once complete, export the video in 1080p.

Conclusion

By following these three phases—Scripting in NotebookLM, Voice Generation in ElevenLabs, and Visual Assembly in Canva/ClipChamp—you can produce high-quality podcast content consistently. The “Zero Cost” nature of this workflow allows you to experiment with different niches and topics until you find an audience, all without spending a single dollar on equipment or software subscriptions.