ElevenLabs and Wondercraft team up to fix NotebookLM podcasts

Google's NotebookLM went viral by turning dry documents into lively two-host podcasts. But anyone who tried using it for serious work hit the same wall: the hosts always sounded the same, you couldn't use your own voice, and making surgical edits was virtually impossible.

Now, audio creation platform Wondercraft and voice giant ElevenLabs are fixing that exact bottleneck. Wondercraft recently introduced full editing support for NotebookLM podcasts, and the traction was immediate: over 10,000 creators imported and edited their AI audio within the first week alone.

Turning generated banter into real shows

Wondercraft is built as an accessible audio studio that requires no sound engineering background. Under the hood, it is powered by ElevenLabs' Voice Library.

With this integration, users can take an raw audio track from NotebookLM and completely overhaul it. By tapping into ElevenLabs' voice cloning tools, creators can swap the generic hosts for their own voices, change accents, fine-tune the tone, or even dial up specific emotional delivery. For fun or specific formats, they can even switch to character voices like pirates or Santa.

So far, users on Wondercraft have generated over 15,000 hours of audio using ElevenLabs voices, representing nearly two years of nonstop playback. More than 10,000 people have already created their own voice clones.

The feature set is also reaching beyond hobbyists. Enterprise clients including Amazon, Siemens, Canon, and Novartis use the platform, alongside high-profile creators like Steven Bartlett, Ali Abdaal, and Lewis Howes. Podcast network sales teams are even utilizing the setup to spin up quick, realistic host-read ad prototypes without dragging talent into the booth.

To prevent abuse, Wondercraft relies on ElevenLabs' "no-go" safeguards, which block attempts to clone public figures and political leaders without authorization, backed by detailed audit logs.

Moving beyond out-of-the-box AI audio

Industry veterans are taking notice. Juergen Berkessel, co-founder of podcast agency Polymash, which has produced over 5,000 episodes, pointed out why this layer matters.

Berkessel noted that he could never publish an out-of-the-box NotebookLM episode for real listeners. However, pairing it with Wondercraft and ElevenLabs changed the equation, allowing him to create synthetic versions of his own voice and confidently recommend the workflow to clients. He added that initial client skepticism toward synthetic voices quickly melted away once they heard the actual fidelity.

Why it matters

We think Wondercraft played this brilliantly. Google proved that the world loves conversational AI audio, but NotebookLM left a massive product gap by offering zero post-production control. Instead of trying to reinvent conversational models from scratch, Wondercraft positioned itself as the logical post-production suite for Google's viral hit.

For independent podcasters and corporate communications teams, this lowers the bar significantly. You can now use NotebookLM to synthesize complex briefs and use Wondercraft to publish polished, branded audio in minutes. For sponsor reads, generating realistic prototypes on the fly will save teams days of studio scheduling.

The broader risk, however, is audio fatigue. As cloning and re-editing become this effortless, podcast feeds risk getting flooded with half-automated shows. While safety checks are holding off celebrity deepfakes, maintaining distinct identity and authentic connection with listeners is going to get a lot harder.