A UK startup that has developed a platform that allows anyone to create a podcast using AI voices and an AI-generated script has raised a $3m seed round.
Wondercraft allows people to create podcasts using synthetic voices from a script written by a user or ChatGPT-like large language model (LLM), as it looks to “empower everyone to be an audio creator”.
It can also be used to produce audiobooks, adverts and corporate training material.
The round was led by US VC Will Ventures, which was joined by Diary of a CEO podcast star Steven Bartlett and and Y Combinator. ElevenLabs, which built the text-to-speech tech Wondercraft runs on and raised an $80m Series B last week, was also involved.
AI audio
Wondercraft launched in beta in May last year and was founded by former Palantir engineers Dimitris Nikolaou and Youssef Rizk and ex-Spotify and Acast executive Oskar Serrander. It’s been used by more than 30k people since then, and officially launched last week.
Using ElevenLabs’ AI, the platform offers a number of text-to-speech templates, ranging from daily news podcasts to briefs-to-audio adverts and audiobooks. Users type a script — or get an AI assistant to write it for them — and select one of 40 synthetic voices to read it.
Creators can also add a sample of their own voices that the text-to-speech model imitates — although when this British Sifted reporter recorded his dulcet tones, the AI spat out a voice with a distinct American lilt.
Jingles can be added into the audio, creating what Serrander calls “studio-quality audio productions”.
Wondercraft also has an AI-driven translation feature — which allows users to dub audio clips into different languages, and is around “90% accurate”, according to Serrander. For premium users — like Steven Bartlett’s Diary of a CEO podcast, which has used the translation feature to dub YouTube videos into Spanish and French — there’s a team of human translators on hand to boost that accuracy.
The ‘middle-class creator’
The startup is targeting two broad groups of users.
The first is the “middle-class creator”, says Serrander. “Just like Canva democratised graphic design and enabled a whole new community of people to become designers and Substack made it easier to share writing — that’s what Wondercraft wants to do for audio.”
Users pay $34 a month for a basic subscription to the platform — which gets you 60 minutes of audio creation and access to 40 voices and voice cloning. The premium service will set you back $109, and provides access to a range of additional features like human translators and the option to clone more than one voice.
Then there’s enterprise clients: bigger podcast networks, media platforms and advertising and streaming services. Wondercraft says it’s onboarding a range of enterprise clients at the moment, and currently works with three global audio networks. Cryptocurrency news site Coindesk, for example, currently uses Wondercraft to generate daily podcasts, says Serrander.
The idea is that it’s hugely expensive to produce audio content like podcasts and adverts, and an AI studio like Wondercraft could cut those costs significantly.
“Every company has a blog, but there’s no way to do that [affordably] with audio. With Wondercraft, you can turn that CEO post into audio,” says Serrander.
But do people really want to listen to AI-generated audio versions of some CEO’s latest musings?
“The staff might want to,” Serrander optimistically replies. “Big companies send out huge long training modules or texts for onboarding and you can make that into audio and way more compelling and get people off the screens a little.”
What’s the market like?
Wondercraft’s existence is in large part due to ElevenLabs’ success.
“The possibility of [making podcasts using synthetic voices] has only been enabled recently due to very realistic text-to-speech models,” says Nikolaou. ElevenLabs, which raised its pre-seed round in January 2023 and provides the tech that Wondercraft runs its app on, has been at the forefront of these developments, he adds.
As generative AI has boomed, many companies have emerged offering AI synthetic voices or dubbing. Alongside ElevenLabs, Descript and WellSaid Labs, from the US, have picked up $100m and $10m from VCs, respectively, according to Dealroom. Murf.ai, in India, has raised $11.5m, and Israel’s Deepdub has raised $20m.
Wondercraft currently has a team of eight, and plans to at least double it in the next 12 months. Most of the money from the raise will go towards building out the engineering team, says Serrander.
