Moises.ai allows AI-generated voices to be detected and created

Music AI company Moises.ai has launched two brand new features that allow artists, labels and entertainment professionals to create and detect end-to-end AI-generated voice synthesis.

AI-powered models can sound just like popular artists, including Drake and the Weeknd, and have been used to create unauthorised new songs.

Moises.ai has developed two new AI-powered models, one of which can detect these synthesised voices and flag them for rightsholders, while the other generates synthetic voices, powered by the sounds of real artists.

Moises.ai has been trusted to help artists, music organisations and labels make the most of their sonic identity since 2019. Now, Moises.ai is expanding this service with synthetic voice detection and voice replication, two different services now necessary for those utilising AI.

Voice replication gives complete end-to-end control over an artist’s AI-generated voice models, offering two levels of control to artists, family or businesses managing a late artist’s estate. Rightsholders can use AI voice detection to find tracks that employ synthesised voices.

With voice replication powered by Moises.ai, rightsholders can require select producers to submit tracks for processing and then see all results submitted before they are distributed or shared publicly.

They could also simply share a model with select partners or producers, granting more creative freedom to generate and release on their own. This places fewer restrictions and hands more creative freedom to the producer, while the rightsholder still retains the ability to see the results of use.

“Moises.ai is uniquely positioned to revolutionise the field of AI voice synthesis, detection and revival, thanks to our acute focus on high fidelity and fine-tuned AI models that deliver the most precise and highest quality output available in the market,” says CEO and co-founder of Moises.ai, Geraldo Ramos.

Moises claims to be one of the only companies focusing on voice synthesis for singing, and unlike other companies, is sourcing its own dataset through partnerships with studios and artists. Using professional audio equipment to record original hi-fi audio at the source, the company says producers will get the highest quality results from their AI model.

Moises.ai also claims to dramatically reduce the time, effort and experience required to engage with AI tools and incorporate them into workflows.

Artists and organisations can now access a hi-fi model in less than 24 hours, providing convenience for those who seek to harness AI voice synthesis or revival for their projects.

Moises.ai has also recently launched a drag-and-drop, no code, DIY AI interface to the public, called Orchestrator, to support stem separation or lyric transcription using AI.

Supported by AI voice synthesis, Moises allows famous characters and artists to perform in multiple languages, creating immersive multilingual experiences for audiences around the globe.

Voice revival also has the power to honour artists who have passed away, by modelling their voices using AI. These models can be used to breathe fresh life into timeless classics, unveil lost treasures, create new works, find new fans, or protect the digital identity of artists’ most precious asset.

“When channelled properly with care and attention, AI voice synthesis can reshape the way media impacts audiences everywhere,” adds Matt Henninger, VP of sales and business development. “Moises.ai works directly with artists, organisations and labels to make sure this tech serves them in all the right ways.”

Moises stresses that it is “always thinking about the rightsholders first” when AI is being used in music and entertainment, including film and television projects.

The new detection and creation features enable rightsholders to find AI tracks that were generated by using synthesised voices without their permission, or to willingly permit the use for monetisation and audience reach possibilities.

These models provide localisation for entertainment professionals in music, television and film to create and detect AI-generated audio.

For example, an English-language video can be dubbed instantly to other languages with the voice and vocal range of the original speaker, enabling the video to be localised elsewhere for wider accessibility and consumption. 

Artists and studios can partner with Moises AI by offering their tracks as a dataset, bringing additional revenue streams.