The AI ​​voice-generating platform shocking the world is getting an update to fight abuse

Generative AI has the ability to generate all types of content including text, art, images and even speech.

AI startup ElevenLabs has supported text-to-speech generation and voice cloning since its beta launch in January and has accumulated over one million registered users.

Too, Meta unveils Voicebox AI to replicate the voices of your friends and loved ones

On Tuesday, ElevenLabs announced The closing of a $19 million Series A round, as well as some major updates to the platform, including addressing one of its biggest controversies.

Since its launch, Elevenlabs’ voice-generating technology has had both positive and negative effects.

Some of the positive uses described by ElevenLabs include, “independent authors creating audiobooks, developers voicing characters in video games, supporting the blind to access written content online, and empowering the world’s first AI radio channel.”

While these use cases are positive and advance the business processes of many different industries, there are equally harmful applications as well.

Voice-cloning tools, which take snippets of a person’s voice to generate new audio, have been used for nefarious means. public figures It sounds like they’re saying horrible, discriminatory statement,

A few weeks after releasing the beta, ElevenLabs quickly took it down Twitter To address “misuse cases of voice cloning”. The company suggested possible ways to deal with this problem such as additional account verification, verifying copyright of voices, taking voice cloning to a paid level, and even manually verifying each request.

Too, Vimeo adds a suite of AI tools to make video creation a whole lot easier

Today, it released to the public what appears to be the company’s solution to the issue, a AI speech classifier, This tool will be able to understand whether the uploaded audio contains AI-generated audio from ElevenLabs.

“The release of the AI ​​Speech Classifier is the company’s latest step in transparency, and a cornerstone of their commitment to creating a secure generative media landscape,” ElevenLabs said in the release.

According to a previous post announcing the tool, the tool maintains 99% accuracy in recognizing audio when it is unmodified.

However, if the audio undergoes codec or reverb transformation, the accuracy drops to over 90% accuracy, and the more the content is processed, the lower the accuracy, release,

This tool will not prevent abuse and may help dispel confusion once the initial damage is done. Its effectiveness in solving the problem is questionable, but it’s a small step.

This is not the first time AI-generation technology has been misused to target public figures. For example, an AI music generator was able to generate a Drake and The Weeknd collaboration that sounded real, even though none of the artists were actually on the track.

Too, Can AI-Generated Music Win Music Awards? Grammy unveils new rules

AI art and image generators have also been used to generate simulated, realistic images of public figures performing certain activities. Some of these images have been used negatively political propaganda While others have been used for entertainment purposes only, such as the meme of Pope Francis in a puffer coat.

In addition to the AI ​​Speech Classifier, ElevenLabs also announced the arrival of “Projects” in its suite of products.

“Projects” is a workflow for editing and creating long-form spoken content, available for early access now. It’s meant to serve as a one-stop shop for audio-editing needs and provide a “Google Docs level of simplicity” for audio creation, according to the release.

The addition of the “Projects” feature is similar to what we’ve seen on other creativity platforms, such as Vimeo, TikTok, and Adobe Express. All of these platforms aim to implement AI in a way that optimizes a user’s workflow and allows for easy, customized creation of content.