How to Produce a Multi-Speaker Podcast Episode in ElevenLabs Studio

Build a scripted two-host podcast where each speaker has a distinct voice, then export the full mix.

10 minIntermediate

Visit ElevenLabsAffiliate link. See our disclosure.

Studio (formerly Projects) is built for long, multi-paragraph content with more than one voice. It is the right tool for a scripted podcast where two hosts trade lines. This guide sets up a project, assigns a voice to each speaker, and exports a finished episode.

What you need

An ElevenLabs plan that includes Studio access
A written script with clear speaker labels, for example Host A and Host B
Two distinct voices chosen or cloned in advance

Step 1: Create a Studio project

Open Studio from the sidebar and click Create new. Give the project an episode title and choose a blank document so you can paste your own script.

Step 2: Paste the script and split paragraphs

Paste your script. Studio breaks the text into paragraphs, each of which can be assigned its own voice. Make sure each host's lines sit in their own paragraph so you can switch voices per block.

ElevenLabs - Studio editor

[ Adam ] So today we are talking about home recording.

[ Rachel ] Right, and the first mistake people make is...

[ Adam ] Exactly. Room treatment matters more than the mic.

----------------------------------------------------------

Default voice: [ Adam v ] Model: [ Multilingual v2 v ]

[ Generate all ]

Each paragraph shows the voice assigned to it on the left.

Step 3: Assign a voice per speaker

Click into each paragraph and set its voice using the per-paragraph voice control. Assign Host A's blocks to one voice and Host B's to another. Keeping the assignment consistent is what makes the dialogue feel like two real people.

Step 4: Tune settings for natural delivery

Adjust stability and similarity for the project or per paragraph. Lower stability gives more expressive, varied delivery; higher stability is steadier but flatter. For conversational podcasts, slightly lower stability usually sounds more lifelike.

Generate one paragraph first

Before generating the whole episode, render a single paragraph to confirm the voice and tone are right. Fixing settings after a full generation wastes characters.

Step 5: Generate the full episode

Once a sample sounds good, generate the entire project. Studio stitches the paragraphs together in order. Listen through and regenerate any individual paragraph that came out wrong without touching the rest.

Step 6: Export the audio

Use the export button to download the finished episode as a single audio file. Bring it into a DAW like Audacity or Reaper if you want to add an intro sting, music bed, or normalization before publishing.

Result: a clean back-and-forth episode with two consistent voices that you can drop into any podcast host.

Ready to try it yourself? Get set up withElevenLabsin a few minutes.(affiliate link. We may earn a commission at no extra cost. Disclosure)