Artificial intelligence capable of creating all types of content including letterart, imageand even words.
AI startup, ElevenLabs, has been supporting text-to-speech and voice transcription since its beta launch in January and has amassed over a million registered users.
Also: Meta reveals Voicebox AI to reproduce the voices of your friends and loved ones
On Tuesday, ElevenLabs announced wraps up a $19 million Series A round, as well as several major updates to the platform, including those to resolve its biggest controversy.
Since its launch, Elevenlabs’ voice-generating technology has had both positive and negative connotations.
Some of the positive uses, as described by ElevenLabs, include “indie authors creating audiobooks, developers voicing video game characters, assisting visually impaired people to access content online written content and power the world’s first AI radio channel.”
While these use cases are positive and drive business processes across various industries, there are equally adverse applications.
The voice transcription engine, which takes fragments of a person’s voice to create new sounds, has been used for nefarious means, causing public figure it sounds like they’re talking terrible, discriminatory statement.
A few weeks after the beta release, ElevenLabs immediately made it Twitter to address “instances of abuse of voice cloning.” The company suggested potential ways to solve the problem, such as additional account verification, voice copyright verification, voice transcript transfer to the paid tier, and even manual verification. each request.
Also: Vimeo adds a suite of AI tools to make video creation significantly easier
Today, it released to the public what appears to be the company’s solution to the problem, a AI voice classifier. This tool will be able to decode whether the uploaded audio contains AI-generated audio from ElevenLabs.
“The release of the AI Speech Classifier is the latest step in the company’s drive to promote transparency, and it’s the cornerstone of their commitment to creating a secure shared communications landscape,” said ElevenLabs. said in the release.
According to a previous post announcing the tool, the tool maintains >99% accuracy in determining when audio has not been modified.
However, if the audio undergoes a Codec or reverb conversion, the accuracy drops to more than 90%, and the more the content is processed, the lower the accuracy will be, according to liberate, release, free.
This tool will not prevent misuse and can only help resolve confusion after the initial harm has been 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 abused to target public figures. For example, an AI music generator can created a Drake and The Weekend collaboration that sounds authentic even though no artist is actually following.
Also: Can AI-generated music win music awards? Grammy reveals new rules
AI art and image maker have also been used to create fake, realistic images of public figures performing certain activities. Some of these images have been used negatively such as political propaganda while others are used for entertainment purposes only, such as the Pope Francis meme in a puffer jacket.
In addition to the AI Voice Classifier, ElevenLabs also announced the arrival of “Project” for its product suite.
“Project” is a workflow for editing and creating long-form spoken content now available for early access. According to the release, it’s meant to serve as a one-stop shop for audio editing needs and offer “Google Docs simplicity” for audio creation.
The addition of the “Projects” feature is similar to what we’ve seen from other creative platforms, such as Vimeo, TikTokAnd Adobe Express. The goal of all these platforms is to implement AI in a way that optimizes user workflow and allows for easier, optimized content creation.
