Respeecher’s ethics-first approach to AI voice cloning locks in new funding

Ukrainian synthetic voice startup Respeecher is finding success despite not just bombs raining down on their city, but a wave of hype that has raised up sometimes controversial competitors. A new $1 million in funding should help the company add a few studios to its media and gaming clients.

Respeecher is perhaps best known for being chosen to replicate James Earl Jones and his iconic Darth Vader voice for a Star Wars animated show, then later a younger Luke Skywalker for The Mandalorian. But the company has also worked with game developer CD Projekt (of Witcher and Cyberpunk fame) and recently wrapped a deal with Warner Music to recreate another iconic voice: French singer Edith Piaf.

Unlike text-to-speech engines, Respeecher uses voice models to modify the speech of actors, who are doing their own best to recreate the voice or character in question. That way it’s not simply generated, but more like a prosthetic voice. They also do accent changing, helpful for reducing an unwanted accent or helping put one on.

The ethical questions involved in cloning someone’s voice are obvious, particularly someone long dead who can’t meaningfully consent. And some startups and services have simply let the cat out of the bag, seeing it as a losing battle in many ways. (Not to mention it limits the range of clientele.)

Respeecher has made ethics a pillar of their business in its various verticals.

“Consent is obtained from those who own the rights; in case of deceased actors, it could be estate or family,” said CEO and co-founder Alex Serdiuk. “There are many cases when they are very involved in the process, and provide valuable feedback to make the voice perfect — as such projects are a tribute to their relatives, their contribution and the characters they built.”

Most recently they worked with Calm to have a voice based on old Hollywood star Jimmy Stewart’s voice.

For the living, permission and compensation are worked out from the start. Voice actors are beginning to see these voice models as assets to control and monetize, instead of (or perhaps in addition to) being a threat to their livelihood. Respeecher is putting together a voice library of actors who have opted in to the process, and the company has also joined Adobe’s Content Authenticity Initiative (for what it’s worth).

By not focusing on scaling like crazy during a big year in AI, Respeecher may have missed out on some capital or business opportunities. But slow and steady may actually serve them well in this case — and besides, they had a lot going on in Kyiv this last year.

“Like all Ukrainian businesses and startups, this war taught us what being resilient really means,” said Serdiuk. Raising funds is never easy, and probably it would be easier if Russia didn’t attack our cities with missiles or Shahid drones regularly. After all this, now, I believe there are hardly any obstacles that our team could not overcome or solutions that we couldn’t find.”

The company has managed to pursue one new vertical during this chaotic time, though: synthetic voices for people who have lost the ability to speak on their own. We’ve seen other startups and established companies entering this space, which may not be as lucrative or flashy but can change lives.

“We have a bunch of projects with hospitals as well as patients with ataxia or laryngectomy. One of the laryngectomy patients we had a chance to work with [is] Konrad Zieliński, a PhD student at the University of Warsaw who had lost his voice due to laryngectomy. Our technology helped him to communicate in a more natural way in his own voice,” said Serdiuk. You can read more about Konrad’s case in this blog post.

Respeecher announced today that it had raised a $1 million “pre-Series A” round, to which entrepreneur Gary Vaynerchuk and funds ffVC Poland, Bad Ideas, ICU and SID Venture Partners contributed.