Named after the Australian mimicking song bird, Lyrebird.ai is working on software that can allegedly recreate someone’s voice based on a few minutes of recorded data.
The company’s aim is to develop a more humanized voicing system for future vocal human-computer interfaces, but of course with technology like this, who knows what the heck can go wrong.
Many voice recorders lack fundamental features like expressing emotion or generating unique intonation patterns. That’s something the company is integrating into its software to bring more realism to its software.
For now the software is still in its infancy, and they are still seeking the help of beta testers to further develop their code, but some of the samples sound pretty decent. You can check out all their demos here, but I pulled one of Barack Obama and Donald Trump having a conversation.
Pretty good for a first build, right?
The scary bit is that the concept works and sets the stage for even more advanced voice duplication software in the future.
Rather than try to defend audio recordings, the company goes straight in to say that we should not trust what we hear on a recording as it’s extremely easy to replicate someone’s voice these days.
Voice recordings are currently considered as strong pieces of evidence in our societies and in particular in jurisdictions of many countries. Our technology questions the validity of such evidence as it allows to easily manipulate audio recordings. This could potentially have dangerous consequences such as misleading diplomats, fraud and more generally any other problem caused by stealing the identity of someone else.
By releasing our technology publicly and making it available to anyone, we want to ensure that there will be no such risks. We hope that everyone will soon be aware that such technology exists and that copying the voice of someone else is possible. More generally, we want to raise attention about the lack of evidence that audio recordings may represent in the near future.
Lyrebird.ai is not the only company working on voice copying software. You may have remembered Adobe’s VoCo demo that allows you to take a recording, and copy/paste segments of it and even generate new sentences.
Looks like the courts may be royally screwed now.