Rephrase.ai raises $1.5M to use synthetic media for personalized sales pitches

Bangalore-based Rephrase.ai has an ambitious vision for reshaping how movies and videos are made.

CEO Ashray Malhotra laid it out for me yesterday, saying that his co-founder Nisheeth Lahoti “came up with this concept — he wants to build an engine that can take any script as input and create a professional movie,” no filming required.

But Rephrase.ai is starting with what Malhotra said is a more “short-term, monetizable” goal: Offering technology that makes it easy to create personalized sales videos.

The startup was part of the Techstars Bangalore program in 2019 and is announcing today that it has raised $1.5 million in seed funding led by Lightspeed Venture Partners and AV8 Ventures.

Malhotra demonstrated the technology for me, showing me how a salesperson can select a model, a background and a voice, and enter text that the model will recite. They can then export that video for use in a variety of sales tools.

This is valuable for, he said, because sending personalized video messages in sales emails can lead to “an insane increase” in clickthrough rates. But creating all those videos can be a huge chore, if not downright impossible.

And while there are plenty of other startups working on synthetic media, Malhotra said Rephrase.ai is set apart by the 18 months the team spent developing technology that can take 10 minutes of footage and “predict how the lip movements of the person would have been if you’d shot them [saying any phrase] in an actual studio.”

https://vimeo.com/459294477

You can see the results for yourself in the video above. Personally, I was impressed by the lip movements but disconcerted by the fact that Rephrase.ai customers can pair any model with any voice, leading to some strange combinations that feel more like badly dubbed movie than an effective sales pitch.

When I brought this up, Malhotra replied that some clients will want to take the time “perfecting it out, finding the right voices, the right costumes, the right personality of the actors,” while other clients might be fine spending less time to create something a little less convincing.

It’s also worth noting that Rephrase.ai has several policies designed to prevent the creation of deceptive deepfakes: Presenters can control who has the authority to create videos using their faces, the platform is only open to authorized businesses and videos are created from scratch, rather than transferring someone’s face onto an existing person.

Malhotra said Rephrase.ai is currently talking to a number of potential customers, but those discussions are in early stages. He also suggested that the technology could expand fairly quickly into areas like chatbots and education.

“I think it’s going to open a whole new world of creativity,” he said. “When you and I want to express something, we’re most likely to write a text document, but as a viewer, we want to see a video. They’ve been disconnected because video creation is really difficult.”

Sonantic is ready to convince listeners that synthetic voices can cry