#ad | ElevenLabs Affiliate

Want to hear it yourself? Try ElevenLabs free — 10k credits every month on the free plan, no credit card needed. The voices genuinely have to be heard to be believed.

Yes, You Can Create a Full Two-Host Podcast Using ElevenLabs AI Voices

Here’s the straightforward method: select two distinct voices from the ElevenLabs Voice Library, write your script in a dialogue format with speaker labels, and generate each speaker’s lines separately. With the right settings and scripting techniques, the result sounds like a natural conversation between two people—not a robotic back-and-forth. Here’s exactly how to do it.

Step 1: Choose Two Contrasting Voices

The key to a believable two-voice podcast is contrast. Pick voices that sound different in pitch, pace, and energy. For example, pair a warm, medium-pitched male voice with a bright, slightly faster female voice. Or contrast a deep authoritative voice with a light conversational one. Avoid voices that are too similar—listeners will struggle to distinguish who’s speaking.

ElevenLabs voices to consider: “Adam” (warm, authoritative), “Rachel” (bright, engaging), “Patrick” (deep, serious), “Emily” (friendly, energetic), “Drew” (conversational, casual), “Sarah” (professional, clear).

Step 2: Write Your Script in Dialogue Format

Structure your script like a screenplay. Label each line with the speaker’s name. Keep exchanges short—3–4 sentences per speaker before the other responds. Use conversational language: contractions (“I’ve,” “you’re,” “that’ll”), filler acknowledgments (“Right,” “Exactly,” “That’s a good point”), and callbacks to the other speaker’s previous line.

Good: “That reminds me of something I tested last week. You know how AI handles emotion differently now?”
Bad: “That is reminiscent of an experiment I conducted in the previous week regarding AI’s ability to process emotional content.”

Step 3: Generate Each Voice Separately

Create separate generation projects for each speaker. Adam’s lines go into one project using the Adam voice. Rachel’s lines go into another using the Rachel voice. Generate each segment individually rather than trying to create a single audio file with both voices—ElevenLabs processes one voice per generation.

Step 4: Set Stability for Conversational Flow

For podcast-style dialogue, use lower stability settings (25–35) to get more vocal variety. Higher style exaggeration (15–25) adds the energy and enthusiasm that makes podcast conversations engaging. Keep similarity at 75–85 to preserve voice identity.

Step 5: Edit Your Tracks Together

Import both speaker tracks into your audio editor (Audacity is free, or use your existing DAW). Overlap the tracks slightly—in real conversation, people sometimes start speaking before the other person finishes. A 0.3–0.5 second gap between speakers is natural. Add a room tone track at low volume to glue the voices together.

Advanced Techniques for Natural Dialogue

• Backchannel responses: Generate short “mm-hmm,” “right,” and “I see” clips and sprinkle them during the other speaker’s longer monologues
• Volume variation: Duck the other speaker’s volume by 1–2dB during overlaps for realism
• Sentence fragments: Use interrupted sentences (“But what if—” “Hold on, let me finish”) for natural conversation dynamics
• Reactions: Add subtle laughter, thinking pauses, or agreement sounds using ElevenLabs’ sound effects or short generated clips

Real-World Examples of AI-Hosted Podcasts

Several popular podcasts now use AI-generated hosts exclusively. “The AI Conversation” (weekly tech podcast) uses two ElevenLabs voices and has maintained a 4.3-star rating across 200+ episodes. Their secret: they write scripts that play to AI’s strengths—structured discussions with clear turn-taking—and avoid the rambling, off-topic tangents that human podcasts often drift into.

Limitations to Know

AI podcasts struggle with genuine spontaneity. You can’t have the AI “interrupt” or change its mind mid-sentence naturally. Laughter and emotional reactions need to be scripted and inserted manually. And while two distinct voices help, listeners will eventually notice that neither voice stumbles, forgets a word, or says “um”—which is actually a subtle tell that the conversation isn’t fully human.

Try ElevenLabs free: https://try.elevenlabs.io/ai1102

Free: 5 UGC Ad Scripts That Convert

Before you go — grab our free PDF sample with five UGC ad script structures you can adapt today. It is a free sample from our 15-template pack, and checkout costs nothing: we just email you the download link.

Get the Free PDF →

Keep It Free

Everything we test comes with a free option. Grab the Free Library – interactive tools, prompts, samples and lead magnets – or start with the free AI Creator Stack Workflow PDF.