Software uses machine-learning to alter voices

Voice-enabled technology transforms user voices to make them sound like a completely different person.

Voice-enabled technology is becoming increasingly popular. Amazon’s virtual assistant Alexa is one of the most popular examples, but Springwise has written about a breadth of innovations in this field. A robot keeps the elderly company and reacts to voice-activated demands. Even board games have had a technological spin, with the world’s first voice-augmented offering being created by a UK startup.

New software created by Boston-based startup Modulate enables users to change their voice as they speak. The software converts a person’s speech into somebody else’s vocal chords. Modulate is working with its primary target market of the gaming industry to produce voice skins. In gaming, skins are aesthetic changes to a weapon or avatar that have no effect on gameplay. Voice skins change the sound of speech without changing the underlying emotion, wording, or overall speech style. The software can also create voices by users entering text, but the voice would consequently lack emotion.

Modulate is aware of the security risk that its technology creates, as the technology can therefore replicate any voice. This could range from politicians to celebrities. The startup says it will not enable voice fraud and it watermarks all the work it creates. If someone feels as if their voice has been misused without their permission, they can contact Modulate to have it removed. The company is accepting requests to join its beta program and will work with early customers to build specific, production-quality voice skins for their services or media. Over the coming months, the company will then be opening up Modulate to more people and releasing the first tools for using voice skins live as a person speaks.