Resemble AI Launches Detector, a Deepfake Identifier

Resemble AI, a provider of generative voice artificial intelligence, today released a deepfake audio detector, Resemble Detect, and Resemble AI Speech-to-Speech voice conversion. The company also just completed an $8 million funding round.

Resemble Detect validates the authenticity of audio data to expose deepfake audio in real time with up to 98 percent accuracy, according to the company. It also analyzes audio across all forms of media and against all modern generative AI speech synthesis solutions.

Earlier this year, Resemble AI released the PerTh Watermarker, which can detect whether a Resemble AI-produced audio file has been manipulated. Now, with Resemble Detect, an additional layer of security is provided to those consuming AI-generated content.

"Our mission is to make interactions with digital products as human and natural as possible," said Zohaib Ahmed, co-founder and CEO of Resemble AI, in a statement. "We want to make AI tools available to more people, but we also acknowledge that we can't talk about AI without talking about ethics. We're proud to release Resemble Detect, another solution in our toolkit to ensure legitimacy."

Resemble Detect uses a state of the art deep neural network trained to identify fakes from real audio. It identifies artifacts—the sonic material that results from the editing or manipulation of sound. It has learned its own version of a spectrogram, allowing it to see different frequencies where artifacts could be contained. This embedding contains specific time and frequency information that allows it to make an optimum prediction between a real and fake audio file. The audio data is then processed and run through a deep learning model, which outputs the likelihood of it being a fake from 0 to 1.

Resemble AI has also made significant improvements to its Speech-to-Speech model, increasing accuracy and robustness, and it is now available to all customers. With real-time voice conversion, natural-sounding AI voices can be created across gaming, entertainment, interactive voice response (IVR), and more. AI voices can perform a wide range of emotions, speaking styles, and languages.