Voice Clone Platform Resemble AI Wins $8 Million

Spread the love

Image credit: Flashpop/Getty Images

Resemble AI, a platform that uses generative AI to clone realistic voices, today announced it has raised $8 million in Series A funding led by Javelin Venture Partners, with participation from Craft Ventures and Ubiquity Ventures. bottom.

The tranche, which brings the startup’s total funding to $12 million, will be used to further develop Resemble’s enterprise offerings and double the size of its team to more than 40 people by the end of the year, said co-founder and CEO. CEO Zohaib Ahmed said.

“Ressemble’s technology is being used by some of the world’s largest media companies to create content that was not possible before,” Ahmed told TechCrunch in an email interview.

Resemble was founded in 2019 by Ahmed Muhammad and Saqib Muhammad who realized that video game audio couldn’t keep up with the frequent version updates of the games themselves. Ahmed had just finished his stint at BlackBerry and his Hipmunk and was working as a lead his software engineer at Magic Leap.

Resemble started small, primarily focused on gaming use cases. However, the platform has grown to offer AI technology that can “transfer” audio to other languages, generate personalized messages from voice actors, and create real-time conversational agents.

Resemble is just one player in the burgeoning speech-generating AI market. Papercup, Deepdub, Eleven Labs, Respeecher, Acapela, and Voice.ai, not to mention big tech incumbents like AWS, Azure, and Google Cloud, offer notable AI tools to clone and generate speech. One of the startup vendors.

However, it’s a controversial technology, and for good reason.

Motherboard notes that voice actors are increasingly being asked to sign the rights to their voices in order to allow clients to leverage AI to ultimately generate a synthetic version that replaces them. writes articles for free.

Image credit: looks like AI

Deepfakes are another matter.

Malicious attackers are using AI to replicate people’s voices, tricking victims into thinking they are talking to relatives or customers. It’s not just the potential for crime that’s ringing alarm bells. In 2021, a documentary filmmaker was accused of hiring a company to clone Anthony Bourdain’s voice after his death, with the consent of Bourdain’s estate. Meanwhile, audio deepfakes have taken the social media by storm, mostly with harmless effects, but sometimes not.

But Ahmed argues that Resemble stands out in the field of ethics.

“In addition to requiring explicit user consent to reproduce audio, strict usage guidelines are applied to prevent malicious use,” he said.

To this end, Resemble requires users to provide a recording of the “consent clip” of the audio they are trying to clone. Resemble blocks users from creating AI voices if the voice in the clip does not match other clips.

Additionally, to prevent abuse when recording, Resemble forces users to say a series of specific sentences aloud. Resemble flags a recording as potentially exploitable if it deviates from the script.

“Once a voice is created, the user owns all rights to that voice,” says Ahmed. “We do not use that voice data to train other models, nor do we resell voice data to third-party companies. We make sure that the voices that the companies are reproducing are available through their voice actors and have the appropriate consents. ”

Resemble also developed Resemble Detect, a product designed to verify the authenticity of speech data using AI models trained to distinguish between real and fake speech. The model essentially “sees” various frequencies that may contain artifacts caused by sound editing and manipulation, and makes predictions about the “reality” of the clip with 0% to 100% confidence. increase.

Detect is intended to complement Resemble’s audio watermarking technology, PerTh Watermark, which uses AI models to generate and insert audio tones imperceptible to the human ear that convey identifying information. (Note that PerTh Watermarker is a kind of platform-locked play. It can only mark and detect Resemble’s own generated voices, not other commercial or open Not compatible with source speech generation AI tools.)

Ahmed credits these tools to Resemble’s success. The platform has more than 1 million users and has generated 35 years’ worth of audio in the last 12 months, he said.

“Regulating AI is a top priority for government officials, so Resemble provides insights and recommendations for responsible use of generated audio,” said Ahmed. “Ressemble makes it easier than ever to create compelling, high-quality audio content, allowing content creators to add a whole new level of credibility to their work, and a new level of authenticity for their audience. It can give you an immersive feeling.”

#Voice #Clone #Platform #Resemble #Wins #Million


Spread the love

Leave a Comment