ElevenLabs is a pacesetter in AI audio. Its instruments, reminiscent of AI voice cloning, have achieved worldwide recognition. At the moment, the startup launched its AI Sound Results instrument to assist creatives discover the proper sound results for his or her initiatives.
Initially introduced in February, the instrument allows you to generate sound results, distinctive character voices, and music snippets from textual content prompts, in accordance with ElevenLabs. You may hear sound results created by the instrument for OpenAI’s Sora demo video beneath:
ElevenLabs says the instruments are supposed to assist folks, together with content material creators, movie and tv studio employees, and online game builders, generate the sounds they should deliver their initiatives to life “affordably and at scale.”
“During the last yr, we have revolutionized AI Voices by producing the primary actually emotive, human-like text-to-speech platform,” ElevenLabs co-founder and CEO Mati Staniszewski stated in a press release. “With the launch of text-to-sound results, we’re marking one other main step ahead, one that may equip creators with extra audio instruments to assist them produce high-quality content material.”
To make AI results attainable, ElevenLabs partnered with Shutterstock to fine-tune its mannequin utilizing content material from the Shutterstock audio library of licensed tracks, addressing moral issues about utilizing a generative AI mannequin.
The AI Sound Results instrument is reside on the ElevenLabs web site, with totally different tiered plans to accommodate person wants. You may attempt the instrument at no cost, though it does depend in direction of your month-to-month 10,000-character restrict.
As somebody who enjoys modifying movies in my spare time and as a part of my job, I used to be enthusiastic about the potential of discovering sound results extra simply. I gave the instrument a attempt to see the way it labored.
To begin, go to the ElevenLabs web site, click on on sound results on the right-hand panel, and kind in what you wish to hear. The primary immediate I typed in was “small canine barking.” The instrument generated 5 totally different variations, as seen beneath:
As a proud Yorkie proprietor, I can attest that the generated sound results had been near the true factor. The instrument was intuitive, and the method was primarily the identical as utilizing most AI picture or music turbines.
After I used a extra complicated immediate, “ladies cheering,” the generator took longer to output a outcome and the standard was not as correct or useable as the primary take a look at. After I returned to easier prompts, nevertheless, reminiscent of “kitchen alarm bell ringing,” I had nice outcomes. The 5 outputs sounded just like the immediate however assorted barely, providing totally different choices.
The AI Sound Results instrument may also generate music. When prompted to create a “lo-fi beat with a jazzy groove,” the instrument produced 5 high-quality choices.
Finally, I used to be impressed with the instrument and encourage you to check it. AI Sound Results is a enjoyable and free expertise. That stated, I’d advocate not asking the instrument to make human sounds. As an alternative, if you wish to generate speech, have a look at ElevenLab’s text-to-speech instrument.